@zhangjian This is a great job, Thanks for working on this. I have a question here. Should we let AsyncHandler thread pool have the ability of isolation? I’m worried about async handler thread will be exhausted when there are some nameservices which have bad performance. | | 张浩博 | | hfutzhan...@163.com |
---- Replied Message ---- | From | zhangjian<1361320...@qq.com.INVALID> | | Date | 06/6/2024 11:49 | | To | Xiaoqiao He<hexiaoq...@apache.org> | | Cc | Hadoop Common<common-dev@hadoop.apache.org> , Hdfs-dev<hdfs-...@hadoop.apache.org> , <priv...@hadoop.apache.org> | | Subject | Re: [Discuss] RBF: Aynchronous router RPC. | Hi, xiaoqiao He, thank you so much! Best Regards, - zhangjian 2024年6月6日 11:33，Xiaoqiao He <hexiaoq...@apache.org> 写道： Hi @zhangjian <1361320...@qq.com> , the dev branch HDFS-17531 is ready now. FYI. Best Regards, - He Xiaoqiao On Tue, Jun 4, 2024 at 10:08 PM zhangjian <1361320...@qq.com> wrote: Hi, Xiaoqiao He: Can you help create a dev branch? I don't have the permission to create it. Thank you very much. - zhangjian 2024年5月30日 11:57，Xiaoqiao He <hexiaoq...@apache.org> 写道： Great! It looks like there are no other nothing blockers. @zhangjian <1361320...@qq.com> If no other furthermore comments, we should go to the next step: a. Create a dev branch for this proposal. b. Split this huge PR to some small JIRA and PRs. c. Involve some folks to review PR. Please ping here if you need any help. Thanks again. Good Luck! Best Regards, - He Xiaoqiao On Wed, May 29, 2024 at 11:46 AM Ayush Saxena <ayush...@gmail.com> wrote: Thanx for the details, sounds cool, good luck with the feature!!! -Ayush On 29 May 2024, at 8:56 AM, zhangjian <1361320...@qq.com> wrote: Thank you for taking the time to review this proposal. Your opinion does point out the key issues to designing an asynchronous router, but my proposal can address these issues: 1. My design does not affect the functionality of existing synchronous routers in throwing stanby or retry exceptions and other aspects, the async router still inherits these implementations. 2. Currently, both asynchronous router and sync router support backpressure on client requests when they exceed a certain limit ( asynchronous router : cannot obtain semaphores through the handler, sync router : block through handler synchronization, unable to obtain available handler ) and return standby exception to allow the client to retry other routers (RouterRpcFairnessPolicyController mechanism). Thank you again! zhangjian 2024年5月29日 07:05，Ayush Saxena <ayush...@gmail.com> 写道： Thanx folks, I had a very quick pass on the PDF and it looks good. Maybe some doubts around the fact where it was mentioned that if a Namenode returns a StandbyException or something on similar lines, the Router will retry, I think we have some logic in RouterRpcClient checking for such case, if it is StandByException it does try the other Namenode, but for all other Retryable Exceptions, we return them back to the client & let the client operate according to its Retry Policy, I think we should preserve that behaviour, if the intentions were to change it. Regarding controlling the concurrency to prevent OOM at the router, maybe we should consider rejecting the client requests beyond a certain limit/backlog & return back a relevant Retriable Exception to the client, so that it can retry on another Router rather than overloading one Router when there are other available, most of the deployments I believe would be running considerable number of Routers Rest I scratched my head for possible scenario where things can go south, but I think mostly the scenarios that came into my mind are covered Nothing blocker from my side, Good Luck!!! -Ayush On Tue, 28 May 2024 at 21:52, Sangjin Lee <sj...@apache.org> wrote: Sounds good. Thanks for sharing your findings. On Sat, May 25, 2024 at 2:24 AM zhangjian <1361320...@qq.com> wrote: Hello everyone, I conducted a performance comparison test between sync and asynchronous router, and the test results showed that in single ns or multi ns scenarios, Asynchronous router in terms of throughput The utilization of CPU and thread, as well as the average processing time of client requests, are better than those of sync router, especially when downstream ns have performance bottlenecks, The performance of the async router is far greater than that of the sync router; And in terms of isolation, Asynchronous router is also better than sync router. Detailed testing PDF: https://issues.apache.org/jira/browse/HDFS-17531 Comparison of Async router & sync router performance.pdf 2024年5月24日 14:13，Yuanbo Liu <liuyuanb...@gmail.com> 写道： good job! On Fri, May 24, 2024 at 1:57 AM zhangjian <1361320...@qq.com> wrote: Hello everyone, currently, I have tested the performance of async and sync router for a downstream ns: 1. The throughput, CPU, and thread performance of the async router are better than those of the sync router, and its memory performance is within an acceptable range compared to the synchronous router. 2. Asynchronous router can apply pressure downstream to better utilize the performance of downstream ns, and can almost fill the call queue of downstream ns. Due to the large size of the test result pdf, it cannot be sent via email, please see: https://issues.apache.org/jira/browse/HDFS-17531 2024年5月23日 17:03，Xiaoqiao He <hexiaoq...@apache.org> 写道： Great. Thanks for your addendum information. cc @Ayush Saxena <ayush...@gmail.com> @inigo...@apache.org <inigo...@apache.org> Any more feedback for this proposal? IMO The feature of asynchronous router RPC is a helpful improvement. For my internal practice, it will improve the throughput of requests forward significantly and is very valuable to push it forward. Thanks again and good luck! Best Regards, - He Xiaoqiao On Wed, May 22, 2024 at 9:59 AM zhangjian <1361320...@qq.com> wrote: Hi, Sangjin Lee, thank you for your attention. I will use my free time to do a performance comparison recently. 2024年5月22日 03:42，Sangjin Lee <sj...@apache.org> 写道： Thanks for the great proposal, Zhangjian. On point #3, I suspect it should be fairly straightforward to create a small isolated synthetic test to prove (or disprove) the benefits of this approach. By driving a controlled amount of requests per second, you could see latency, memory, CPU, etc. Ideally, it should show meaningful improvements without much degradation in other metrics. Would you be able to spend some time doing that? Thanks, Sangjin On Tue, May 21, 2024 at 5:13 AM zhangjian <1361320...@qq.com.invalid> wrote: Hi, xiaoqiao he, thank you for your reply. 1.Currently, the server and client protocols within router can be implemented by extends existing protocols and adding asynchronous functionality, so it will not affect existing synchronization protocols. RouterClientNamenodeProtocolServerSideTranslatorPB RouterClientProtocolTranslatorPB RouterGetUserMappingsProtocolServerSideTranslatorPB RouterGetUserMappingsProtocolTranslatorPB RouterNamenodeProtocolServerSideTranslatorPB RouterNamenodeProtocolTranslatorPB RouterRefreshUserMappingsProtocolServerSideTranslatorPB RouterRefreshUserMappingsProtocolTranslatorPB The following issues have implemented asynchronous callbacks for Rpc.server, but I have not found any other modules to use related functions Server HADOOP-11552 HADOOP-17046 In the implementation of asynchronous Rpc.client, this issue is directly used Client HADOOP-13226 Therefore, I believe that asynchronous routers are safe for modifying the RPC protocol, RPC server, and client 2. Forwarding requests to multiple downstream ns, the synchronous router handler adds requests from multiple downstream ns to the thread pool (RouterRpcClient.executorService), and then waits for responses from all downstream ns before returning. Since threads in the thread pool also process rpc requests synchronously, similar to a handler, the number of threads in the thread pool directly affects the performance of invoiceConcurrent, which in turn affects the performance of the handler. In asynchronous router implementation, the handler calls invoiceConcurrent to simply convert a request into multiple requests and add them to the asyn handler thread pool, which can then process the next request in the call queue; When a connection thread of a downstream ns receives a response, it will hand it over to the async response for processing. The async response thread will determine whether it has received all responses from the downstream ns. If it does, it will continue to process the response. Otherwise, the async response thread will process the next response. The asynchronous router uses CompletableFuture.allOf() to implement asynchronous invoiceConcurrent, and the handler, async handler, async response, and connection thread still does not need to wait synchronously. In addition, synchronous routers not only have drawbacks in multi ns environments, but also in single downstream ns situations, it is often difficult to decide how many handlers to set for the router, setting it too much will waste thread resources, and setting it too small will not be able to give pressure to downstream ns; Asynchronous routers can push requests to downstream ns without considering how to set handlers. Asynchronous routers can also better connect to more downstream storage services that support the HDFS protocol, with better scalability. 3.Since I have not yet deployed asynchronous routers to our own cluster, there is no performance comparison. However, theoretically, I believe that asynchronous routers will occupy more memory than synchronous routers. However, I do not believe that it will occupy a lot, especially since we can control the maximum number of requests entering the router, as CompletableFuture is stable and widely used; In other aspects, it should be far superior to synchronous routers, especially in downstream scenarios with more ns.If anyone is interested, you can also help to make a performance comparison 2024年5月21日 11:39，Xiaoqiao He <hexiaoq...@apache.org> 写道： Thanks for this great proposal! Some questions after reviewing the design doc (sorry didn't review PR carefully which is too large.) 1. This solution will involve RPC framework update, will it affect other modules and how to keep other modules off these changes. 2. Some RPC requests should be forward concurrently to all downstream NS, will it cover this case in this solution. 3. Considering there is one init-version implementation, did you collect some benchmark vs the current synchronous model of DFSRouter? Thanks again. Best Regards, - He Xiaoqiao On Tue, May 21, 2024 at 11:21 AM zhangjian <1361320...@qq.com.invalid> wrote: Thank you for your positive attitude towards this feature. You can debug the UTs provided in PR to better understand the current asynchronous calling function. 2024年5月21日 02:04，Simbarashe Dzinamarira < simbadz...@apache.org> 写道： Excited to see this feature as well. I'll spend more time understanding the proposal and implementation. On Mon, May 20, 2024 at 7:55 AM zhangjian <1361320...@qq.com.invalid wrote: Hi, Yuanbo liu, thank you for your interest in this feature, I think the difficulty of an asynchronous router is not only to implement asynchronous functions, but also to consider the readability and reusability of the code, so as to facilitate the development of the community. I also planned to do the virtual thread you mentioned at the beginning, virtual Threads can achieve asynchronousization elegantly at the code level, but the biggest problem is that it is not easy to upgrade the jdk version, no matter in the community or in the actual production environment. Therefore, I later used CompletableFuture, which is currently supported by jdk 8, to achieve asynchronousization. The router is stateless, and the router rpc process is very clear. Therefore, even if CompletableFuture itself is not as readable as the virtual thread, if we design it well, we can make the asynchronous process look very clear. 2024年5月20日 10:56，Yuanbo Liu <liuyuanb...@gmail.com> 写道： Nice to see this feature brought up. I tried to implement this feature in our internal clusters, and know that it's a very complicated feature, CC hdfs-dev to bring more discussion. By the way, I'm not sure whether virtual thread of higher jdk will help in this case. On Mon, May 20, 2024 at 10:10 AM zhangjian <1361320...@qq.com.invalid wrote: Hello everyone, currently there are some shortcomings in the RPC of HDFS router： Currently the router's handler thread is synchronized, when the *handler* thread adds the call to connection.calls, it needs to wait until the *connection* notifies the call to complete, and then Only after the response is put into the response queue can a new call be obtained from the call queue and processed. Therefore, the concurrency performance of the router is limited by the number of handlers; a simple example is as follows: If the number of handlers is 1 and the maximum number of calls in the connection thread is 10, then even if the connection thread can send 10 requests to the downstream ns, since the number of handlers is 1, the router can only process one request after another. Since the performance of router rpc is mainly limited by the number of handlers, the most effective way to improve rpc performance currently is to increase the number of handlers. Letting the router create a large number of handler threads will also increase the number of thread switches and cannot maximize the use of machine performance. There are usually multiple ns downstream of the router. If the handler forwards the request to an ns with poor performance, it will cause the handler to wait for a long time. Due to the reduction of available handlers, the router's ability to handle ns requests with normal performance will be reduced. From the perspective of the client, the performance of the downstream ns of the router has deteriorated at this time. We often find that the call queue of the downstream ns is not high, but the call queue of the router is very high. Therefore, although the main function of the router is to federate and handle requests from multiple NSs, the current synchronous RPC performance cannot satisfy the scenario where there are many NSs downstream of the router. Even if the concurrent performance of the router can be improved by increasing the number of handlers, it is still relatively slow. More threads will increase the CPU context switching time, and in fact many of the handler threads are in a blocked state, which is undoubtedly a waste of thread resources. When a request enters the router, there is no guarantee that there will be a running handler at this time. Therefore, I consider asynchronous router rpc. Please view the issues: https://issues.apache.org/jira/browse/HDFS-17531 for the complete solution. And you can also view this PR: https://github.com/apache/hadoop/pull/6838, which is just a demo, but it completes the core asynchronous RPC function. If you think asynchronous routing is feasible, we can consider splitting this PR for easy review in the future. The PDF is attached and can also be viewed through issues. Welcome everyone to exchange and discuss! --------------------------------------------------------------------- To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-dev-h...@hadoop.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-dev-h...@hadoop.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-dev-h...@hadoop.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: common-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: common-dev-h...@hadoop.apache.org
