Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
These are both really good posts: you should try and get them in to the documentation. with anything implementing dynamicness, there are some fun problems (a) detecting the delays in the workflow. There's some good ideas here (b) deciding where to address it. That means you need to monitor the entire pipeline —which you should be doing in production anyway. (c) choosing the action. More nodes, more memory CPU (not that useful for Java code, even when YARN adds support for dynamic container resize) (d) choosing the size of the action. In a shared cluster, extra resources for one app comes at the expense of others. If you have pre-emption turned on in YARN, the scheduler can take containers off lower priority work, which automates a lot of this decision making. That will lose other work though, so to justify it you'd better hang on those containers (e) deciding if/when to hand things back. Scaling things down can be very expensive if lots of state has to get rebuilt elsewhere. I think Apache Helix from LinkedIn has done some good work here -worth looking at to see what lessons code to lift. And as you'd expect, it sits right behind Kafka in production. I think it gets away with low delays to scale up/down and relying on low rebuild costs. [In the work I've been doing with colleagues on dynamic HBase and Accumulo clusters, we've not attempted to do any autoscale, because scale down is an expensive decision...we're focusing on liveness detection and reaction, then publishing the metrics needed to allow people or cross-application tools to make the decision) On 12 Jun 2015, at 04:38, Dmitry Goldenberg dgoldenberg...@gmail.commailto:dgoldenberg...@gmail.com wrote: Yes, Tathagata, thank you. For #1, the 'need detection', one idea we're entertaining is timestamping the messages coming into the Kafka topics. The consumers would check the interval between the time they get the message and that message origination timestamp. As Kafka topics start to fill up more, we would presumably see longer and longer wait times (delays) for messages to be getting processed by the consumers. The consumers would then start firing off critical events into an Event Analyzer/Aggregator which would decide that more resources are needed, then ask the Provisioning Component to allocate N new machines. We do want to set maxRatePerPartition in order to not overwhelm the consumers and run out of memory. Machine provisioning may take a while, and if left with no maxRate guards, our consumers could run out of memory. Since there are no receivers, if the cluster gets a new executor, it will automatically start getting used to run tasks... no need to do anything further. This is great, actually. We were wondering whether we'd need to restart the consumers once the new machines have been added. Tathagata's point implies, as I read it, that no further orchestration is needed, the load will start getting redistributed automatically. This makes implementation of autoscaling a lot simpler, as far as #3. One issue that's not yet been covered much is the scenario when *fewer* cluster resources become required (a system load valley rather than a peak). To detect a low volume, we'd need to measure the throughput in messages per second over time. Real low volumes would cause firing off of critical events signaling to the Analyzer that machines could be decommissioned. If machines are being decommissioned, it would seem that the consumers would need to get acquiesced (allowed to process any current batch, then shut down), then they would restart themselves or be restarted. Thoughts on this? There is also a hefty #4 here which is the hysteresis of this, where the system operates adaptively and learns over time, remembering the history of cluster expansions and contractions and allowing a certain slack for letting things cool down or heat up more gradually; also not contracting or expanding too frequently. PID controllers and thermostat types of design patterns have been mentioned before in this discussion. If you look at the big cloud apps, they dynamically reallocate VM images based on load history, with Netflix being the poster user: Hadoop work in the quiet hours, user interaction evenings and weekends. Excluding special events (including holidays), there's a lot of regularity over time, which lets you predict workload in advance. It's like your thermostat knowing fridays are cold and it should crank up the heating in advance. On Thu, Jun 11, 2015 at 11:08 PM, Tathagata Das t...@databricks.commailto:t...@databricks.com wrote: Let me try to add some clarity in the different thought directions that's going on in this thread. 1. HOW TO DETECT THE NEED FOR MORE CLUSTER RESOURCES? If there are not rate limits set up, the most reliable way to detect whether the current Spark cluster is being insufficient to handle the data load is to use the StreamingListner interface which
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
If I want to restart my consumers into an updated cluster topology after the cluster has been expanded or contracted, would I need to call stop() on them, then call start() on them, or would I need to instantiate and start new context objects (new JavaStreamingContext(...)) ? I'm thinking of actually acquiescing these streaming consumers but letting them finish their current batch first. Right now I'm doing jssc.start(); jssc.awaitTermination(); Must jssc.close() be called as well, after awaitTermination(), to avoid potentially leaking contexts? I don't see that in things like JavaDirectKafkaWordCount but wondering if that's needed. On Wed, Jun 3, 2015 at 11:49 AM, Evo Eftimov evo.efti...@isecc.com wrote: Makes sense especially if you have a cloud with “infinite” resources / nodes which allows you to double, triple etc in the background/parallel the resources of the currently running cluster I was thinking more about the scenario where you have e.g. 100 boxes and want to / can add e.g. 20 more *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:46 PM *To:* Evo Eftimov *Cc:* Cody Koeninger; Andrew Or; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, One of the ideas is to shadow the current cluster. This way there's no extra latency incurred due to shutting down of the consumers. If two sets of consumers are running, potentially processing the same data, that is OK. We phase out the older cluster and gradually flip over to the new one, insuring no downtime or extra latency. Thoughts? On Wed, Jun 3, 2015 at 11:27 AM, Evo Eftimov evo.efti...@isecc.com wrote: You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” You should be able to bring new worker nodes online and make them contact and register with the Master without bringing down the Master (or any of the currently running worker nodes) Then just shutdown your currently running spark streaming job/app and restart it with new params to take advantage of the larger cluster *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:14 PM *To:* Cody Koeninger *Cc:* Andrew Or; Evo Eftimov; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Let me try to add some clarity in the different thought directions that's going on in this thread. 1. HOW TO DETECT THE NEED FOR MORE CLUSTER RESOURCES? If there are not rate limits set up, the most reliable way to detect whether the current Spark cluster is being insufficient to handle the data load is to use the StreamingListner interface which gives all the information about when batches start and end. See the internal implementation of the StreamingListener called StreamingJobProgressListener. This is the one that drives the streaming UI. You can get the scheduling delay (time take for a batch to start processing) from it and use that as a reliable indicator that Spark Streaming is not able to process as fast as data is being received. But if you have already set rate limits based on the max load that cluster can handle, then you will probably never detect that the actual input rate into Kafka has gone up and data is getting buffered inside Kafka. In that case, you have to monitor kafka load to correctly detect the high load. You may to use a combination of both techniques for robust and safe elastic solution - Have rate limits set, use StreamingListener for early detect that processing load is increasing (can increase without actual increase in data rate) and also make sure from Kafka monitoring that the whole end-to-end system is keeping up. 2. HOW TO GET MORE CLUSTER RESOURCES? Currently for YARN, you can use the developer API of dynamic allocation that Andrew Or has introduced to ask for more executors from YARN. Note that the existing dynamic allocation solution is unlikely to work for streaming, and should not be used. Rather I recommend building your own logic that sees the streaming scheduling delay, and accordingly uses the low level developer API to directly ask for more executors (sparkContext.requestExecutors). In other approaches, the Provising Component idea can also work. 3. HOW TO TAKE ADVANTAGE OF MORE CLUSTER RESOURCES? There are two approaches depending on receiver vs Kafka direct. I am assuming the number of topic partitions pre-determined to be large enough to handle peak load. (a) Kafka Direct: This is the simpler scenario. Since there are no receivers, if the cluster gets a new executor, it will automatically start getting used to run tasks, including reading from Kafka (remember, Kafka direct approach reads from Kafka like a file system, from any node that runs the task). So it will immediately start using the extra resources, no need to do anything further. (b) Receiver: This is definitely tricky. If you dont need to increase the number of receivers, then a new executor will start getting used for computations (shuffles, writing out, etc.), but the parallelism in receiving will not increase. If you need to increase that, then its best to shutdown the context gracefully (so that no data is lost), and a new StreamingContext can be started with more receivers (# receivers = # executors), and may be more #partitions for shuffles. You have call stop on currently running streaming context, to start a new one. If a context is stopped, any thread stuck in awaitTermniation will get unblocked. Does that clarify things? On Thu, Jun 11, 2015 at 7:30 AM, Cody Koeninger c...@koeninger.org wrote: Depends on what you're reusing multiple times (if anything). Read http://spark.apache.org/docs/latest/programming-guide.html#rdd-persistence On Wed, Jun 10, 2015 at 12:18 AM, Dmitry Goldenberg dgoldenberg...@gmail.com wrote: At which point would I call cache()? I just want the runtime to spill to disk when necessary without me having to know when the necessary is. On Thu, Jun 4, 2015 at 9:42 AM, Cody Koeninger c...@koeninger.org wrote: direct stream isn't a receiver, it isn't required to cache data anywhere unless you want it to. If you want it, just call cache. On Thu, Jun 4, 2015 at 8:20 AM, Dmitry Goldenberg dgoldenberg...@gmail.com wrote: set the storage policy for the DStream RDDs to MEMORY AND DISK - it appears the storage level can be specified in the createStream methods but not createDirectStream... On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Yes, Tathagata, thank you. For #1, the 'need detection', one idea we're entertaining is timestamping the messages coming into the Kafka topics. The consumers would check the interval between the time they get the message and that message origination timestamp. As Kafka topics start to fill up more, we would presumably see longer and longer wait times (delays) for messages to be getting processed by the consumers. The consumers would then start firing off critical events into an Event Analyzer/Aggregator which would decide that more resources are needed, then ask the Provisioning Component to allocate N new machines. We do want to set maxRatePerPartition in order to not overwhelm the consumers and run out of memory. Machine provisioning may take a while, and if left with no maxRate guards, our consumers could run out of memory. Since there are no receivers, if the cluster gets a new executor, it will automatically start getting used to run tasks... no need to do anything further. This is great, actually. We were wondering whether we'd need to restart the consumers once the new machines have been added. Tathagata's point implies, as I read it, that no further orchestration is needed, the load will start getting redistributed automatically. This makes implementation of autoscaling a lot simpler, as far as #3. One issue that's not yet been covered much is the scenario when *fewer* cluster resources become required (a system load valley rather than a peak). To detect a low volume, we'd need to measure the throughput in messages per second over time. Real low volumes would cause firing off of critical events signaling to the Analyzer that machines could be decommissioned. If machines are being decommissioned, it would seem that the consumers would need to get acquiesced (allowed to process any current batch, then shut down), then they would restart themselves or be restarted. Thoughts on this? There is also a hefty #4 here which is the hysteresis of this, where the system operates adaptively and learns over time, remembering the history of cluster expansions and contractions and allowing a certain slack for letting things cool down or heat up more gradually; also not contracting or expanding too frequently. PID controllers and thermostat types of design patterns have been mentioned before in this discussion. On Thu, Jun 11, 2015 at 11:08 PM, Tathagata Das t...@databricks.com wrote: Let me try to add some clarity in the different thought directions that's going on in this thread. 1. HOW TO DETECT THE NEED FOR MORE CLUSTER RESOURCES? If there are not rate limits set up, the most reliable way to detect whether the current Spark cluster is being insufficient to handle the data load is to use the StreamingListner interface which gives all the information about when batches start and end. See the internal implementation of the StreamingListener called StreamingJobProgressListener. This is the one that drives the streaming UI. You can get the scheduling delay (time take for a batch to start processing) from it and use that as a reliable indicator that Spark Streaming is not able to process as fast as data is being received. But if you have already set rate limits based on the max load that cluster can handle, then you will probably never detect that the actual input rate into Kafka has gone up and data is getting buffered inside Kafka. In that case, you have to monitor kafka load to correctly detect the high load. You may to use a combination of both techniques for robust and safe elastic solution - Have rate limits set, use StreamingListener for early detect that processing load is increasing (can increase without actual increase in data rate) and also make sure from Kafka monitoring that the whole end-to-end system is keeping up. 2. HOW TO GET MORE CLUSTER RESOURCES? Currently for YARN, you can use the developer API of dynamic allocation that Andrew Or has introduced to ask for more executors from YARN. Note that the existing dynamic allocation solution is unlikely to work for streaming, and should not be used. Rather I recommend building your own logic that sees the streaming scheduling delay, and accordingly uses the low level developer API to directly ask for more executors (sparkContext.requestExecutors). In other approaches, the Provising Component idea can also work. 3. HOW TO TAKE ADVANTAGE OF MORE CLUSTER RESOURCES? There are two approaches depending on receiver vs Kafka direct. I am assuming the number of topic partitions pre-determined to be large enough to handle peak load. (a) Kafka Direct: This is the simpler scenario. Since there are no receivers, if the cluster gets a new executor, it will automatically start getting used to run tasks, including reading from Kafka (remember, Kafka direct approach reads from Kafka like a file system, from any node that runs the task). So it will immediately start using the extra resources, no need to
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
At which point would I call cache()? I just want the runtime to spill to disk when necessary without me having to know when the necessary is. On Thu, Jun 4, 2015 at 9:42 AM, Cody Koeninger c...@koeninger.org wrote: direct stream isn't a receiver, it isn't required to cache data anywhere unless you want it to. If you want it, just call cache. On Thu, Jun 4, 2015 at 8:20 AM, Dmitry Goldenberg dgoldenberg...@gmail.com wrote: set the storage policy for the DStream RDDs to MEMORY AND DISK - it appears the storage level can be specified in the createStream methods but not createDirectStream... On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
set the storage policy for the DStream RDDs to MEMORY AND DISK - it appears the storage level can be specified in the createStream methods but not createDirectStream... On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not about scaling your system/solution In terms of how to scale the Spark Framework Dynamically – even though this is not supported at the moment out of the box I guess you can have a sys management framework spin dynamically a few more boxes (spark worker nodes), stop dynamically your currently running Spark Streaming Job, relaunch it with new params e.g. more Receivers, larger number of Partitions (hence tasks), more RAM per executor
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
direct stream isn't a receiver, it isn't required to cache data anywhere unless you want it to. If you want it, just call cache. On Thu, Jun 4, 2015 at 8:20 AM, Dmitry Goldenberg dgoldenberg...@gmail.com wrote: set the storage policy for the DStream RDDs to MEMORY AND DISK - it appears the storage level can be specified in the createStream methods but not createDirectStream... On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not about scaling your system/solution In terms of how to scale the Spark Framework Dynamically – even though this is not supported at the moment out of the box I guess you can have a sys management framework spin
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Great. You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” -- anything specific you were thinking of? On Wed, Jun 3, 2015 at 11:49 AM, Evo Eftimov evo.efti...@isecc.com wrote: Makes sense especially if you have a cloud with “infinite” resources / nodes which allows you to double, triple etc in the background/parallel the resources of the currently running cluster I was thinking more about the scenario where you have e.g. 100 boxes and want to / can add e.g. 20 more *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:46 PM *To:* Evo Eftimov *Cc:* Cody Koeninger; Andrew Or; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, One of the ideas is to shadow the current cluster. This way there's no extra latency incurred due to shutting down of the consumers. If two sets of consumers are running, potentially processing the same data, that is OK. We phase out the older cluster and gradually flip over to the new one, insuring no downtime or extra latency. Thoughts? On Wed, Jun 3, 2015 at 11:27 AM, Evo Eftimov evo.efti...@isecc.com wrote: You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” You should be able to bring new worker nodes online and make them contact and register with the Master without bringing down the Master (or any of the currently running worker nodes) Then just shutdown your currently running spark streaming job/app and restart it with new params to take advantage of the larger cluster *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:14 PM *To:* Cody Koeninger *Cc:* Andrew Or; Evo Eftimov; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion with TD (Spark streaming lead) about what needs to be done to provide some semblance of dynamic scaling to streaming applications, e.g. take into account the batch queue instead. We came up with a
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
If we have a hand-off between the older consumer and the newer consumer, I wonder if we need to manually manage the offsets in Kafka so as not to miss some messages as the hand-off is happening. Or if we let the new consumer run for a bit then let the old consumer know the 'new guy is in town' then the old consumer can be shut off. Some overlap is OK... On Wed, Jun 3, 2015 at 11:49 AM, Evo Eftimov evo.efti...@isecc.com wrote: Makes sense especially if you have a cloud with “infinite” resources / nodes which allows you to double, triple etc in the background/parallel the resources of the currently running cluster I was thinking more about the scenario where you have e.g. 100 boxes and want to / can add e.g. 20 more *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:46 PM *To:* Evo Eftimov *Cc:* Cody Koeninger; Andrew Or; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, One of the ideas is to shadow the current cluster. This way there's no extra latency incurred due to shutting down of the consumers. If two sets of consumers are running, potentially processing the same data, that is OK. We phase out the older cluster and gradually flip over to the new one, insuring no downtime or extra latency. Thoughts? On Wed, Jun 3, 2015 at 11:27 AM, Evo Eftimov evo.efti...@isecc.com wrote: You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” You should be able to bring new worker nodes online and make them contact and register with the Master without bringing down the Master (or any of the currently running worker nodes) Then just shutdown your currently running spark streaming job/app and restart it with new params to take advantage of the larger cluster *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:14 PM *To:* Cody Koeninger *Cc:* Andrew Or; Evo Eftimov; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion with TD (Spark streaming lead) about what needs to be done to provide some semblance of dynamic scaling to streaming applications, e.g. take into account the batch queue instead. We came up with a few ideas that I will not detail here, but we are looking into this and do intend to support it in the near future. -Andrew 2015-05-28 8:02 GMT-07:00 Evo Eftimov evo.efti...@isecc.com: Probably you should ALWAYS keep the RDD storage policy to MEMORY AND DISK – it will be your insurance policy against sys crashes due to memory leaks. Until there is free RAM, spark streaming (spark) will NOT resort to disk – and of course resorting to disk from time to time (ie when there is no free RAM ) and taking a performance hit from that, BUT only until there is no free RAM *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Thursday, May 28, 2015 2:34 PM *To:* Evo Eftimov *Cc:* Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, good points. On the dynamic resource allocation, I'm surmising this only works within a particular cluster setup. So it improves the usage of current cluster resources but it doesn't make the cluster itself elastic. At least, that's my understanding. Memory + disk would be good and hopefully it'd take *huge* load on the system to start exhausting the disk space too. I'd guess that falling onto disk will make things significantly slower due to the extra I/O. Perhaps we'll really want all of these elements eventually. I think we'd want to start with memory only, keeping maxRate low enough not to overwhelm the consumers; implement the cluster autoscaling. We might experiment with dynamic resource allocation before we get to implement the cluster autoscale. On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the
RE: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Makes sense especially if you have a cloud with “infinite” resources / nodes which allows you to double, triple etc in the background/parallel the resources of the currently running cluster I was thinking more about the scenario where you have e.g. 100 boxes and want to / can add e.g. 20 more From: Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] Sent: Wednesday, June 3, 2015 4:46 PM To: Evo Eftimov Cc: Cody Koeninger; Andrew Or; Gerard Maas; spark users Subject: Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, One of the ideas is to shadow the current cluster. This way there's no extra latency incurred due to shutting down of the consumers. If two sets of consumers are running, potentially processing the same data, that is OK. We phase out the older cluster and gradually flip over to the new one, insuring no downtime or extra latency. Thoughts? On Wed, Jun 3, 2015 at 11:27 AM, Evo Eftimov evo.efti...@isecc.com wrote: You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” You should be able to bring new worker nodes online and make them contact and register with the Master without bringing down the Master (or any of the currently running worker nodes) Then just shutdown your currently running spark streaming job/app and restart it with new params to take advantage of the larger cluster From: Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] Sent: Wednesday, June 3, 2015 4:14 PM To: Cody Koeninger Cc: Andrew Or; Evo Eftimov; Gerard Maas; spark users Subject: Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion with TD (Spark streaming lead) about what needs to be done to provide some semblance of dynamic scaling to streaming applications, e.g. take into account the batch queue instead. We came up with a few ideas that I will not detail here, but we are looking into this and do intend to support it in the near future. -Andrew 2015-05-28 8:02 GMT-07:00 Evo Eftimov evo.efti...@isecc.com: Probably you should ALWAYS keep the RDD storage policy to MEMORY
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
I think what we'd want to do is track the ingestion rate in the consumer(s) via Spark's aggregation functions and such. If we're at a critical level (load too high / load too low) then we issue a request into our Provisioning Component to add/remove machines. Once it comes back with an OK, each consumer can finish its current batch, then terminate itself, and restart with a new context. The new context would be aware of the updated cluster - correct? Therefore the refreshed consumer would restart on the updated cluster. Could we even terminate the consumer immediately upon sensing a critical event? When it would restart, could it resume right where it left off? On Wed, Jun 3, 2015 at 11:49 AM, Evo Eftimov evo.efti...@isecc.com wrote: Makes sense especially if you have a cloud with “infinite” resources / nodes which allows you to double, triple etc in the background/parallel the resources of the currently running cluster I was thinking more about the scenario where you have e.g. 100 boxes and want to / can add e.g. 20 more *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:46 PM *To:* Evo Eftimov *Cc:* Cody Koeninger; Andrew Or; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, One of the ideas is to shadow the current cluster. This way there's no extra latency incurred due to shutting down of the consumers. If two sets of consumers are running, potentially processing the same data, that is OK. We phase out the older cluster and gradually flip over to the new one, insuring no downtime or extra latency. Thoughts? On Wed, Jun 3, 2015 at 11:27 AM, Evo Eftimov evo.efti...@isecc.com wrote: You should monitor vital performance / job clogging stats of the Spark Streaming Runtime not “kafka topics” You should be able to bring new worker nodes online and make them contact and register with the Master without bringing down the Master (or any of the currently running worker nodes) Then just shutdown your currently running spark streaming job/app and restart it with new params to take advantage of the larger cluster *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Wednesday, June 3, 2015 4:14 PM *To:* Cody Koeninger *Cc:* Andrew Or; Evo Eftimov; Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Would it be possible to implement Spark autoscaling somewhat along these lines? -- 1. If we sense that a new machine is needed, by watching the data load in Kafka topic(s), then 2. Provision a new machine via a Provisioner interface (e.g. talk to AWS and get a machine); 3. Create a shadow/mirror Spark master running alongside the initial version which talks to N machines. The new mirror version is aware of N+1 machines (or N+M if we had decided we needed M new boxes). 4. The previous version of the Spark runtime is acquiesced/decommissioned. We possibly get both clusters working on the same data which may actually be OK (at least for our specific use-cases). 5. Now the new Spark cluster is running. Similarly, the decommissioning of M unused boxes would happen, via this notion of a mirror Spark runtime. How feasible would it be for such a mirrorlike setup to be created, especially created programmatically? Especially point #3. The other idea we'd entertained was to bring in a new machine, acquiesce down all currently running workers by telling them to process their current batch then shut down, then restart the consumers now that Spark is aware of a modified cluster. This has the drawback of a downtime that may not be tolerable in terms of latency, by the system's clients waiting for their responses in a synchronous fashion. Thanks. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Which would imply that if there was a load manager type of service, it could signal to the driver(s) that they need to acquiesce, i.e. process what's at hand and terminate. Then bring up a new machine, then restart the driver(s)... Same deal with removing machines from the cluster. Send a signal for the drivers to pipe down and terminate, then restart them. On Thu, May 28, 2015 at 5:15 PM, Cody Koeninger c...@koeninger.org wrote: I'm not sure that points 1 and 2 really apply to the kafka direct stream. There are no receivers, and you know at the driver how big each of your batches is. On Thu, May 28, 2015 at 2:21 PM, Andrew Or and...@databricks.com wrote: Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion with TD (Spark streaming lead) about what needs to be done to provide some semblance of dynamic scaling to streaming applications, e.g. take into account the batch queue instead. We came up with a few ideas that I will not detail here, but we are looking into this and do intend to support it in the near future. -Andrew 2015-05-28 8:02 GMT-07:00 Evo Eftimov evo.efti...@isecc.com: Probably you should ALWAYS keep the RDD storage policy to MEMORY AND DISK – it will be your insurance policy against sys crashes due to memory leaks. Until there is free RAM, spark streaming (spark) will NOT resort to disk – and of course resorting to disk from time to time (ie when there is no free RAM ) and taking a performance hit from that, BUT only until there is no free RAM *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Thursday, May 28, 2015 2:34 PM *To:* Evo Eftimov *Cc:* Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, good points. On the dynamic resource allocation, I'm surmising this only works within a particular cluster setup. So it improves the usage of current cluster resources but it doesn't make the cluster itself elastic. At least, that's my understanding. Memory + disk would be good and hopefully it'd take *huge* load on the system to start exhausting the disk space too. I'd guess that falling onto disk will make things significantly slower due to the extra I/O. Perhaps we'll really want all of these elements eventually. I think we'd want to start with memory only, keeping maxRate low enough not to overwhelm the consumers; implement the cluster autoscaling. We might experiment with dynamic resource allocation before we get to implement the cluster autoscale. On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From:
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Hi all, As the author of the dynamic allocation feature I can offer a few insights here. Gerard's explanation was both correct and concise: dynamic allocation is not intended to be used in Spark streaming at the moment (1.4 or before). This is because of two things: (1) Number of receivers is necessarily fixed, and these are started in executors. Since we need a receiver for each InputDStream, if we kill these receivers we essentially stop the stream, which is not what we want. It makes little sense to close and restart a stream the same way we kill and relaunch executors. (2) Records come in every batch, and when there is data to process your executors are not idle. If your idle timeout is less than the batch duration, then you'll end up having to constantly kill and restart executors. If your idle timeout is greater than the batch duration, then you'll never kill executors. Long answer short, with Spark streaming there is currently no straightforward way to scale the size of your cluster. I had a long discussion with TD (Spark streaming lead) about what needs to be done to provide some semblance of dynamic scaling to streaming applications, e.g. take into account the batch queue instead. We came up with a few ideas that I will not detail here, but we are looking into this and do intend to support it in the near future. -Andrew 2015-05-28 8:02 GMT-07:00 Evo Eftimov evo.efti...@isecc.com: Probably you should ALWAYS keep the RDD storage policy to MEMORY AND DISK – it will be your insurance policy against sys crashes due to memory leaks. Until there is free RAM, spark streaming (spark) will NOT resort to disk – and of course resorting to disk from time to time (ie when there is no free RAM ) and taking a performance hit from that, BUT only until there is no free RAM *From:* Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] *Sent:* Thursday, May 28, 2015 2:34 PM *To:* Evo Eftimov *Cc:* Gerard Maas; spark users *Subject:* Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, good points. On the dynamic resource allocation, I'm surmising this only works within a particular cluster setup. So it improves the usage of current cluster resources but it doesn't make the cluster itself elastic. At least, that's my understanding. Memory + disk would be good and hopefully it'd take *huge* load on the system to start exhausting the disk space too. I'd guess that falling onto disk will make things significantly slower due to the extra I/O. Perhaps we'll really want all of these elements eventually. I think we'd want to start with memory only, keeping maxRate low enough not to overwhelm the consumers; implement the cluster autoscaling. We might experiment with dynamic resource allocation before we get to implement the cluster autoscale. On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka
Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Hi, tl;dr At the moment (with a BIG disclaimer *) elastic scaling of spark streaming processes is not supported. *Longer version.* I assume that you are talking about Spark Streaming as the discussion is about handing Kafka streaming data. Then you have two things to consider: the Streaming receivers and the Spark processing cluster. Currently, the receiving topology is static. One receiver is allocated with each DStream instantiated and it will use 1 core in the cluster. Once the StreamingContext is started, this topology cannot be changed, therefore the number of Kafka receivers is fixed for the lifetime of your DStream. What we do is to calculate the cluster capacity and use that as a fixed upper bound (with a margin) for the receiver throughput. There's work in progress to add a reactive model to the receiver, where backpressure can be applied to handle overload conditions. See https://issues.apache.org/jira/browse/SPARK-7398 Once the data is received, it will be processed in a 'classical' Spark pipeline, so previous posts on spark resource scheduling might apply. Regarding metrics, the standard metrics subsystem of spark will report streaming job performance. Check the driver's metrics endpoint to peruse the available metrics: driver:ui-port/metrics/json -kr, Gerard. (*) Spark is a project that moves so fast that statements might be invalidated by new work every minute. On Thu, May 28, 2015 at 1:21 AM, dgoldenberg dgoldenberg...@gmail.com wrote: Hi, I'm trying to understand if there are design patterns for autoscaling Spark (add/remove slave machines to the cluster) based on the throughput. Assuming we can throttle Spark consumers, the respective Kafka topics we stream data from would start growing. What are some of the ways to generate the metrics on the number of new messages and the rate they are piling up? This perhaps is more of a Kafka question; I see a pretty sparse javadoc with the Metric interface and not much else... What are some of the ways to expand/contract the Spark cluster? Someone has mentioned Mesos... I see some info on Spark metrics in the Spark monitoring guide https://spark.apache.org/docs/latest/monitoring.html . Do we want to perhaps implement a custom sink that would help us autoscale up or down based on the throughput? -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Autoscaling-Spark-cluster-based-on-topic-sizes-rate-of-growth-in-Kafka-or-Spark-s-metrics-tp23062.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Thank you, Gerard. We're looking at the receiver-less setup with Kafka Spark streaming so I'm not sure how to apply your comments to that case (not that we have to use receiver-less but it seems to offer some advantages over the receiver-based). As far as the number of Kafka receivers is fixed for the lifetime of your DStream -- this may be OK to start with. What I'm researching is the ability to add worker nodes to the Spark cluster when needed and remove them when no longer needed. Do I understand correctly that a single receiver may cause work to be farmed out to multiple 'slave' machines/worker nodes? If that's the case, we're less concerned with multiple receivers; we're concerned with the worker node cluster itself. If we use the ConsumerOffsetChecker class in Kafka that Rajesh mentioned and instrument dynamic adding/removal of machines, my subsequent questions then are, a) will Spark sense the addition of a new node / is it sufficient that the cluster manager is aware, then work just starts flowing there? and b) what would be a way to gracefully remove a worker node when the load subsides, so that no currently running Spark job is killed? - Dmitry On Thu, May 28, 2015 at 7:36 AM, Gerard Maas gerard.m...@gmail.com wrote: Hi, tl;dr At the moment (with a BIG disclaimer *) elastic scaling of spark streaming processes is not supported. *Longer version.* I assume that you are talking about Spark Streaming as the discussion is about handing Kafka streaming data. Then you have two things to consider: the Streaming receivers and the Spark processing cluster. Currently, the receiving topology is static. One receiver is allocated with each DStream instantiated and it will use 1 core in the cluster. Once the StreamingContext is started, this topology cannot be changed, therefore the number of Kafka receivers is fixed for the lifetime of your DStream. What we do is to calculate the cluster capacity and use that as a fixed upper bound (with a margin) for the receiver throughput. There's work in progress to add a reactive model to the receiver, where backpressure can be applied to handle overload conditions. See https://issues.apache.org/jira/browse/SPARK-7398 Once the data is received, it will be processed in a 'classical' Spark pipeline, so previous posts on spark resource scheduling might apply. Regarding metrics, the standard metrics subsystem of spark will report streaming job performance. Check the driver's metrics endpoint to peruse the available metrics: driver:ui-port/metrics/json -kr, Gerard. (*) Spark is a project that moves so fast that statements might be invalidated by new work every minute. On Thu, May 28, 2015 at 1:21 AM, dgoldenberg dgoldenberg...@gmail.com wrote: Hi, I'm trying to understand if there are design patterns for autoscaling Spark (add/remove slave machines to the cluster) based on the throughput. Assuming we can throttle Spark consumers, the respective Kafka topics we stream data from would start growing. What are some of the ways to generate the metrics on the number of new messages and the rate they are piling up? This perhaps is more of a Kafka question; I see a pretty sparse javadoc with the Metric interface and not much else... What are some of the ways to expand/contract the Spark cluster? Someone has mentioned Mesos... I see some info on Spark metrics in the Spark monitoring guide https://spark.apache.org/docs/latest/monitoring.html . Do we want to perhaps implement a custom sink that would help us autoscale up or down based on the throughput? -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Autoscaling-Spark-cluster-based-on-topic-sizes-rate-of-growth-in-Kafka-or-Spark-s-metrics-tp23062.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
RE: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
@DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not about scaling your system/solution In terms of how to scale the Spark Framework Dynamically – even though this is not supported at the moment out of the box I guess you can have a sys management framework spin dynamically a few more boxes (spark worker nodes), stop dynamically your currently running Spark Streaming Job, relaunch it with new params e.g. more Receivers, larger number of Partitions (hence tasks), more RAM per executor etc. Obviously this will cause some temporary delay in fact interruption in your processing but if the business use case can tolerate that then go for it From: Gerard Maas [mailto:gerard.m...@gmail.com] Sent: Thursday, May 28, 2015 12:36 PM To: dgoldenberg Cc: spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Hi, tl;dr At the moment (with a BIG disclaimer *) elastic scaling of spark streaming processes is not supported. Longer version. I assume that you are talking about Spark Streaming as the discussion is about handing Kafka streaming data. Then you have two things to consider: the Streaming receivers and the Spark processing cluster. Currently, the receiving topology is static. One receiver is allocated with each DStream instantiated and it will use 1 core in the cluster. Once the StreamingContext is started, this topology cannot be changed, therefore the number of Kafka receivers is fixed for the lifetime of your DStream. What we do is to calculate the cluster capacity and use that as a fixed upper bound (with a margin) for the receiver throughput. There's work in progress to add a reactive model to the receiver, where backpressure can be applied to handle overload conditions. See https://issues.apache.org/jira/browse/SPARK-7398 Once the data is received, it will be processed in a 'classical' Spark pipeline, so previous posts on spark resource scheduling might apply. Regarding metrics, the standard metrics subsystem of spark will report streaming job performance. Check the driver's metrics endpoint to peruse the available metrics: driver:ui-port/metrics/json -kr, Gerard. (*) Spark is a project that moves so fast that statements might be invalidated by new work every minute. On Thu, May 28, 2015 at 1:21 AM, dgoldenberg dgoldenberg...@gmail.com wrote: Hi, I'm trying to understand if there are design patterns for autoscaling Spark (add/remove slave machines to the cluster) based on the throughput. Assuming we can throttle Spark consumers, the respective Kafka topics we stream data from would start growing. What are some of the ways to generate the metrics on the number of new messages and the rate they are piling up? This perhaps is more of a Kafka question; I see a pretty sparse javadoc with the Metric interface and not much else... What are some of the ways to expand/contract the Spark cluster? Someone has mentioned Mesos... I see some info on Spark metrics in the Spark monitoring guide https://spark.apache.org/docs/latest/monitoring.html . Do we want to perhaps implement a custom sink that would help us autoscale up or down based on the throughput? -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Autoscaling-Spark-cluster-based-on-topic-sizes-rate-of-growth-in-Kafka-or-Spark-s-metrics-tp23062.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail:
Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not about scaling your system/solution In terms of how to scale the Spark Framework Dynamically – even though this is not supported at the moment out of the box I guess you can have a sys management framework spin dynamically a few more boxes (spark worker nodes), stop dynamically your currently running Spark Streaming Job, relaunch it with new params e.g. more Receivers, larger number of Partitions (hence tasks), more RAM per executor etc. Obviously this will cause some temporary delay in fact interruption in your processing but if the business use case can tolerate that then go for it *From:* Gerard Maas [mailto:gerard.m...@gmail.com] *Sent:* Thursday, May 28, 2015 12:36 PM *To:* dgoldenberg *Cc:* spark users *Subject:* Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Hi, tl;dr At the moment (with a BIG disclaimer *) elastic scaling of spark streaming processes is not supported. *Longer version.* I assume that you are talking about Spark Streaming as the discussion is about handing Kafka streaming data. Then you have two things to consider: the Streaming receivers and the Spark processing cluster. Currently, the receiving topology is static. One receiver is allocated with each DStream instantiated and it will use 1 core in the cluster. Once the StreamingContext is started, this topology cannot be changed, therefore the number of Kafka receivers is fixed for the lifetime of your DStream. What we do is to calculate the cluster capacity and use that as a fixed upper bound (with a margin) for the receiver throughput. There's work in progress to add a reactive model to the receiver, where backpressure can be applied to handle overload conditions. See https://issues.apache.org/jira/browse/SPARK-7398 Once the data is received, it will be processed in a 'classical' Spark pipeline, so previous posts on spark resource scheduling might apply. Regarding metrics, the standard metrics subsystem of spark will report streaming job performance. Check the driver's metrics endpoint to peruse the available metrics: driver:ui-port/metrics/json -kr, Gerard. (*) Spark is a project that moves so fast that statements might be
Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile div Original message /divdivFrom: Dmitry Goldenberg dgoldenberg...@gmail.com /divdivDate:2015/05/28 13:16 (GMT+00:00) /divdivTo: Evo Eftimov evo.efti...@isecc.com /divdivCc: Gerard Maas gerard.m...@gmail.com,spark users user@spark.apache.org /divdivSubject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? /divdiv /divThanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently and benefit from ongoing online learning – BUT this is STILL about voluntarily sacrificing your performance in the name of keeping your system stable – it is not about scaling your system/solution In terms of how to scale the Spark Framework Dynamically – even though this is not supported at the moment out of the box I guess you can have a sys management framework spin dynamically a few more boxes (spark worker nodes), stop dynamically your currently running Spark Streaming Job, relaunch it with new params e.g. more Receivers, larger number of Partitions (hence tasks), more RAM per executor etc. Obviously this will cause some temporary delay in fact interruption in your processing but if the business use case can tolerate that then go for it From: Gerard Maas [mailto:gerard.m...@gmail.com] Sent: Thursday, May 28, 2015 12:36 PM To: dgoldenberg Cc: spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Hi, tl;dr At the moment (with a BIG disclaimer *) elastic scaling of spark streaming processes is not supported. Longer version. I assume that you are talking about Spark Streaming as the discussion is about handing Kafka streaming data. Then you have two things to consider: the Streaming receivers and the Spark processing cluster. Currently, the receiving topology is static. One receiver is allocated with each DStream instantiated and it will use 1 core in the cluster. Once the StreamingContext is started, this topology cannot be changed, therefore the number of Kafka receivers
Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Evo, good points. On the dynamic resource allocation, I'm surmising this only works within a particular cluster setup. So it improves the usage of current cluster resources but it doesn't make the cluster itself elastic. At least, that's my understanding. Memory + disk would be good and hopefully it'd take *huge* load on the system to start exhausting the disk space too. I'd guess that falling onto disk will make things significantly slower due to the extra I/O. Perhaps we'll really want all of these elements eventually. I think we'd want to start with memory only, keeping maxRate low enough not to overwhelm the consumers; implement the cluster autoscaling. We might experiment with dynamic resource allocation before we get to implement the cluster autoscale. On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in previous postings submitted by me You can even go one step further and periodically issue “get/check free memory” to see whether it is decreasing relentlessly at a constant rate – if it touches a predetermined RAM threshold that should be your third metric Re the “back pressure” mechanism – this is a Feedback Loop mechanism and you can implement one on your own without waiting for Jiras and new features whenever they might be implemented by the Spark dev team – moreover you can avoid using slow mechanisms such as ZooKeeper and even incorporate some Machine Learning in your Feedback Loop to make it handle the message consumption rate more intelligently
RE: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Probably you should ALWAYS keep the RDD storage policy to MEMORY AND DISK – it will be your insurance policy against sys crashes due to memory leaks. Until there is free RAM, spark streaming (spark) will NOT resort to disk – and of course resorting to disk from time to time (ie when there is no free RAM ) and taking a performance hit from that, BUT only until there is no free RAM From: Dmitry Goldenberg [mailto:dgoldenberg...@gmail.com] Sent: Thursday, May 28, 2015 2:34 PM To: Evo Eftimov Cc: Gerard Maas; spark users Subject: Re: FW: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Evo, good points. On the dynamic resource allocation, I'm surmising this only works within a particular cluster setup. So it improves the usage of current cluster resources but it doesn't make the cluster itself elastic. At least, that's my understanding. Memory + disk would be good and hopefully it'd take *huge* load on the system to start exhausting the disk space too. I'd guess that falling onto disk will make things significantly slower due to the extra I/O. Perhaps we'll really want all of these elements eventually. I think we'd want to start with memory only, keeping maxRate low enough not to overwhelm the consumers; implement the cluster autoscaling. We might experiment with dynamic resource allocation before we get to implement the cluster autoscale. On Thu, May 28, 2015 at 9:05 AM, Evo Eftimov evo.efti...@isecc.com wrote: You can also try Dynamic Resource Allocation https://spark.apache.org/docs/1.3.1/job-scheduling.html#dynamic-resource-allocation Also re the Feedback Loop for automatic message consumption rate adjustment – there is a “dumb” solution option – simply set the storage policy for the DStream RDDs to MEMORY AND DISK – when the memory gets exhausted spark streaming will resort to keeping new RDDs on disk which will prevent it from crashing and hence loosing them. Then some memory will get freed and it will resort back to RAM and so on and so forth Sent from Samsung Mobile Original message From: Evo Eftimov Date:2015/05/28 13:22 (GMT+00:00) To: Dmitry Goldenberg Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? You can always spin new boxes in the background and bring them into the cluster fold when fully operational and time that with job relaunch and param change Kafka offsets are mabaged automatically for you by the kafka clients which keep them in zoomeeper dont worry about that ad long as you shut down your job gracefuly. Besides msnaging the offsets explicitly is not a big deal if necessary Sent from Samsung Mobile Original message From: Dmitry Goldenberg Date:2015/05/28 13:16 (GMT+00:00) To: Evo Eftimov Cc: Gerard Maas ,spark users Subject: Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Thanks, Evo. Per the last part of your comment, it sounds like we will need to implement a job manager which will be in control of starting the jobs, monitoring the status of the Kafka topic(s), shutting jobs down and marking them as ones to relaunch, scaling the cluster up/down by adding/removing machines, and relaunching the 'suspended' (shut down) jobs. I suspect that relaunching the jobs may be tricky since that means keeping track of the starter offsets in Kafka topic(s) from which the jobs started working on. Ideally, we'd want to avoid a re-launch. The 'suspension' and relaunching of jobs, coupled with the wait for the new machines to come online may turn out quite time-consuming which will make for lengthy request times, and our requests are not asynchronous. Ideally, the currently running jobs would continue to run on the machines currently available in the cluster. In the scale-down case, the job manager would want to signal to Spark's job scheduler not to send work to the node being taken out, find out when the last job has finished running on the node, then take the node out. This is somewhat like changing the number of cylinders in a car engine while the car is running... Sounds like a great candidate for a set of enhancements in Spark... On Thu, May 28, 2015 at 7:52 AM, Evo Eftimov evo.efti...@isecc.com wrote: @DG; The key metrics should be - Scheduling delay – its ideal state is to remain constant over time and ideally be less than the time of the microbatch window - The average job processing time should remain less than the micro-batch window - Number of Lost Jobs – even if there is a single Job lost that means that you have lost all messages for the DStream RDD processed by that job due to the previously described spark streaming memory leak condition and subsequent crash – described in
Re: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
bq. detect the presence of a new node and start utilizing it My understanding is that Spark is concerned with managing executors. Whether request for an executor is fulfilled on an existing node or a new node is up to the underlying cluster manager (YARN e.g.). Assuming the cluster is single tenant, executor requests are more likely to be fulfilled on the new nodes. Please correct me if I am wrong. On Wed, May 27, 2015 at 8:26 PM, Dmitry Goldenberg dgoldenberg...@gmail.com wrote: Thanks, Rajesh. I think that acquring/relinquishing executors is important but I feel like there are at least two layers for resource allocation and autoscaling. It seems that acquiring and relinquishing executors is a way to optimize resource utilization within a pre-set Spark cluster of machines. However, to accommodate for big spikes in input data, we also need the actual cluster scaling, i.e. adding (or removing, when no longer needed) worker node machines automatically. On that front, I wonder how Spark reacts to a machine being added or removed and what the actual procedure would be. If we're running on a Hadoop cluster, there's a description of adding a node http://wiki.apache.org/hadoop/FAQ#I_have_a_new_node_I_want_to_add_to_a_running_Hadoop_cluster.3B_how_do_I_start_services_on_just_one_node.3F there. There's also discussions of Hadoop node adding/removal such as this one http://stackoverflow.com/questions/16774439/how-do-i-correctly-remove-nodes-in-hadoop . My worry is, will Spark gracefully and quickly detect the presence of a new node and start utilizing it (i.e. how much does it communicate with the Hadoop cluster manager?)... By the same token, if a node is removed, how can it be removed gracefully so as not to affect/kill any running Spark jobs? On Wed, May 27, 2015 at 10:57 PM, rajesh_kall...@dellteam.com wrote: *Dell - Internal Use - Confidential * Did you check https://drive.google.com/file/d/0B7tmGAdbfMI2OXl6azYySk5iTGM/edit and http://spark.apache.org/docs/latest/job-scheduling.html#dynamic-resource-allocation Not sure if the spark kafka receiver emits metrics on the lag, check this link out http://community.spiceworks.com/how_to/77610-how-far-behind-is-your-kafka-consumer You should be able to whip up a script that runs the Kafka ConsumerOffsetChecker periodically and pipe it to a metrics backend of your choice. Based on this you can work the dynamic resource allocation magic. -Original Message- From: dgoldenberg [mailto:dgoldenberg...@gmail.com] Sent: Wednesday, May 27, 2015 6:21 PM To: user@spark.apache.org Subject: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Hi, I'm trying to understand if there are design patterns for autoscaling Spark (add/remove slave machines to the cluster) based on the throughput. Assuming we can throttle Spark consumers, the respective Kafka topics we stream data from would start growing. What are some of the ways to generate the metrics on the number of new messages and the rate they are piling up? This perhaps is more of a Kafka question; I see a pretty sparse javadoc with the Metric interface and not much else... What are some of the ways to expand/contract the Spark cluster? Someone has mentioned Mesos... I see some info on Spark metrics in the Spark monitoring guide . Do we want to perhaps implement a custom sink that would help us autoscale up or down based on the throughput? -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Autoscaling-Spark-cluster-based-on-topic-sizes-rate-of-growth-in-Kafka-or-Spark-s-metrics-tp23062.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
RE: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics?
Dell - Internal Use - Confidential Did you check https://drive.google.com/file/d/0B7tmGAdbfMI2OXl6azYySk5iTGM/edit and http://spark.apache.org/docs/latest/job-scheduling.html#dynamic-resource-allocation Not sure if the spark kafka receiver emits metrics on the lag, check this link out http://community.spiceworks.com/how_to/77610-how-far-behind-is-your-kafka-consumer You should be able to whip up a script that runs the Kafka ConsumerOffsetChecker periodically and pipe it to a metrics backend of your choice. Based on this you can work the dynamic resource allocation magic. -Original Message- From: dgoldenberg [mailto:dgoldenberg...@gmail.com] Sent: Wednesday, May 27, 2015 6:21 PM To: user@spark.apache.org Subject: Autoscaling Spark cluster based on topic sizes/rate of growth in Kafka or Spark's metrics? Hi, I'm trying to understand if there are design patterns for autoscaling Spark (add/remove slave machines to the cluster) based on the throughput. Assuming we can throttle Spark consumers, the respective Kafka topics we stream data from would start growing. What are some of the ways to generate the metrics on the number of new messages and the rate they are piling up? This perhaps is more of a Kafka question; I see a pretty sparse javadoc with the Metric interface and not much else... What are some of the ways to expand/contract the Spark cluster? Someone has mentioned Mesos... I see some info on Spark metrics in the Spark monitoring guide . Do we want to perhaps implement a custom sink that would help us autoscale up or down based on the throughput? -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Autoscaling-Spark-cluster-based-on-topic-sizes-rate-of-growth-in-Kafka-or-Spark-s-metrics-tp23062.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
