Re: Optimisation advice for Avro->Parquet merge job
Hey Kiran,
Thanks very much for the response. I left for vacation before I could try
this out, but I'll experiment once I get back and let you know how it goes.
Thanks!
James.
On 8 June 2015 at 12:34, kiran lonikar wrote:
> It turns out my assumption on load and unionAll being blocking is not
> correct. They are transformations. So instead of just running only the load
> and unionAll in the run() methods, I think you will have to save the
> intermediate dfInput[i] to temp (parquet) files (possibly to in memory DFS
> like http://tachyon-project.org/) in the run() methods. The second for
> loop will also have to load from the intermediate parquet files. Then
> finally save the final dfInput[0] to the HDFS.
>
> I think this way of parallelizing will force the cluster to utilize the
> all the resources.
>
> -Kiran
>
> On Mon, Jun 8, 2015 at 12:30 PM, kiran lonikar wrote:
>
>> James,
>>
>> As I can see, there are three distinct parts to your program:
>>
>>- for loop
>>- synchronized block
>>- final outputFrame.save statement
>>
>> Can you do a separate timing measurement by putting a simple
>> System.currentTimeMillis() around these blocks to know how much they are
>> taking and then try to optimize where it takes longest? In the second
>> block, you may want to measure the time for the two statements. Improving
>> this boils down to playing with spark settings.
>>
>> Now consider the first block: I think this is a classic case of merge
>> sort or a reduce tree. You already tried to improve this by submitting jobs
>> in parallel using executor pool/Callable etc.
>>
>> To further improve the parallelization, I suggest you use a reduce tree
>> like approach. For example, lets say you want to compute sum of all
>> elements of an array in parallel. The way its solved for a GPU like
>> platform is you divide your input array initially in chunks of 2, compute
>> those n/2 sums parallely on separate threads and save the result in the
>> first of the two elements. In the next iteration, you compute n/4 sums
>> parallely of the earlier sums and so on till you are left with only two
>> elements whose sum gives you final sum.
>>
>> You are performing many sequential unionAll operations for inputs.size()
>> avro files. Assuming the unionAll() on DataFrame is blocking (and not a
>> simple transformation like on RDDs) and actually performs the union
>> operation, you will certainly benefit by parallelizing this loop. You may
>> change the loop to something like below:
>>
>> // pseudo code only
>> int n = inputs.size()
>> // initialize executor
>> executor = new FixedThreadPoolExecutor(n/2)
>> dfInput = new DataFrame[n/2]
>> for(int i =0;i < n/2;i++) {
>> executor.submit(new Runnable() {
>> public void run() {
>> // union of i and i+n/2
>> // showing [] only to bring out array access. Replace with
>> dfInput(i) in your code
>> dfInput[i] = sqlContext.load(inputPaths.get(i),
>> "com.databricks.spark.avro").unionAll(sqlContext.load(inputsPath.get(i +
>> n/2), "com.databricks.spark.avro"))
>> }
>> });
>> }
>>
>> executor.awaitTermination(0, TimeUnit.SECONDS)
>>
>> int steps = log(n)/log(2.0)
>> for(s = 2; s < steps;s++) {
>> int stride = n/(1 << s); // n/(2^s)
>> for(int i = 0;i < stride;i++) {
>> executor.submit(new Runnable() {
>> public void run() {
>> // union of i and i+n/2
>> // showing [] only to bring out array access. Replace
>> with dfInput(i) and dfInput(i+stride) in your code
>> dfInput[i] = dfInput[i].unionAll(dfInput[i + stride])
>> }
>> });
>> }
>> executor.awaitTermination(0, TimeUnit.SECONDS)
>> }
>>
>> Let me know if it helped.
>>
>> -Kiran
>>
>>
>> On Thu, Jun 4, 2015 at 8:57 PM, James Aley
>> wrote:
>>
>>> Thanks for the confirmation! We're quite new to Spark, so a little
>>> reassurance is a good thing to have sometimes :-)
>>>
>>> The thing that's concerning me at the moment is that my job doesn't seem
>>> to run any faster with more compute resources added to the cluster, and
>>> this is proving a little tricky to debug. There are a lot of variables, so
>>> here's what we've tried already and the apparent impact. If anyone has any
>>> further suggestions, we'd love to hear!
>>>
>>> * Increase the "minimum" number of output files (targetPartitions
>>> above), so that input groups smaller than our minimum chunk size can still
>>> be worked on by more than one executor. This does measurably speed things
>>> up, but obviously it's a trade-off, as the original goal for this job is to
>>> merge our data into fewer, larger files.
>>>
>>> * Submit many jobs in parallel, by running the above code in a Callable,
>>> on an executor pool. This seems to help, to some extent, but I'm not sure
>>> what else needs to be configured alongside it -- driver threads, scheduling
>>> policy, etc. We set scheduling to "FAIR" when doing this, as that
Re: Optimisation advice for Avro->Parquet merge job
It turns out my assumption on load and unionAll being blocking is not
correct. They are transformations. So instead of just running only the load
and unionAll in the run() methods, I think you will have to save the
intermediate dfInput[i] to temp (parquet) files (possibly to in memory DFS
like http://tachyon-project.org/) in the run() methods. The second for loop
will also have to load from the intermediate parquet files. Then finally
save the final dfInput[0] to the HDFS.
I think this way of parallelizing will force the cluster to utilize the all
the resources.
-Kiran
On Mon, Jun 8, 2015 at 12:30 PM, kiran lonikar wrote:
> James,
>
> As I can see, there are three distinct parts to your program:
>
>- for loop
>- synchronized block
>- final outputFrame.save statement
>
> Can you do a separate timing measurement by putting a simple
> System.currentTimeMillis() around these blocks to know how much they are
> taking and then try to optimize where it takes longest? In the second
> block, you may want to measure the time for the two statements. Improving
> this boils down to playing with spark settings.
>
> Now consider the first block: I think this is a classic case of merge sort
> or a reduce tree. You already tried to improve this by submitting jobs in
> parallel using executor pool/Callable etc.
>
> To further improve the parallelization, I suggest you use a reduce tree
> like approach. For example, lets say you want to compute sum of all
> elements of an array in parallel. The way its solved for a GPU like
> platform is you divide your input array initially in chunks of 2, compute
> those n/2 sums parallely on separate threads and save the result in the
> first of the two elements. In the next iteration, you compute n/4 sums
> parallely of the earlier sums and so on till you are left with only two
> elements whose sum gives you final sum.
>
> You are performing many sequential unionAll operations for inputs.size()
> avro files. Assuming the unionAll() on DataFrame is blocking (and not a
> simple transformation like on RDDs) and actually performs the union
> operation, you will certainly benefit by parallelizing this loop. You may
> change the loop to something like below:
>
> // pseudo code only
> int n = inputs.size()
> // initialize executor
> executor = new FixedThreadPoolExecutor(n/2)
> dfInput = new DataFrame[n/2]
> for(int i =0;i < n/2;i++) {
> executor.submit(new Runnable() {
> public void run() {
> // union of i and i+n/2
> // showing [] only to bring out array access. Replace with
> dfInput(i) in your code
> dfInput[i] = sqlContext.load(inputPaths.get(i),
> "com.databricks.spark.avro").unionAll(sqlContext.load(inputsPath.get(i +
> n/2), "com.databricks.spark.avro"))
> }
> });
> }
>
> executor.awaitTermination(0, TimeUnit.SECONDS)
>
> int steps = log(n)/log(2.0)
> for(s = 2; s < steps;s++) {
> int stride = n/(1 << s); // n/(2^s)
> for(int i = 0;i < stride;i++) {
> executor.submit(new Runnable() {
> public void run() {
> // union of i and i+n/2
> // showing [] only to bring out array access. Replace with
> dfInput(i) and dfInput(i+stride) in your code
> dfInput[i] = dfInput[i].unionAll(dfInput[i + stride])
> }
> });
> }
> executor.awaitTermination(0, TimeUnit.SECONDS)
> }
>
> Let me know if it helped.
>
> -Kiran
>
>
> On Thu, Jun 4, 2015 at 8:57 PM, James Aley
> wrote:
>
>> Thanks for the confirmation! We're quite new to Spark, so a little
>> reassurance is a good thing to have sometimes :-)
>>
>> The thing that's concerning me at the moment is that my job doesn't seem
>> to run any faster with more compute resources added to the cluster, and
>> this is proving a little tricky to debug. There are a lot of variables, so
>> here's what we've tried already and the apparent impact. If anyone has any
>> further suggestions, we'd love to hear!
>>
>> * Increase the "minimum" number of output files (targetPartitions above),
>> so that input groups smaller than our minimum chunk size can still be
>> worked on by more than one executor. This does measurably speed things up,
>> but obviously it's a trade-off, as the original goal for this job is to
>> merge our data into fewer, larger files.
>>
>> * Submit many jobs in parallel, by running the above code in a Callable,
>> on an executor pool. This seems to help, to some extent, but I'm not sure
>> what else needs to be configured alongside it -- driver threads, scheduling
>> policy, etc. We set scheduling to "FAIR" when doing this, as that seemed
>> like the right approach, but we're not 100% confident. It seemed to help
>> quite substantially anyway, so perhaps this just needs further tuning?
>>
>> * Increasing executors, RAM, etc. This doesn't make a difference by
>> itself for this job, so I'm thinking we're already not fully utilising the
>> resources we have in a smaller cluste
Re: Optimisation advice for Avro->Parquet merge job
James,
As I can see, there are three distinct parts to your program:
- for loop
- synchronized block
- final outputFrame.save statement
Can you do a separate timing measurement by putting a simple
System.currentTimeMillis() around these blocks to know how much they are
taking and then try to optimize where it takes longest? In the second
block, you may want to measure the time for the two statements. Improving
this boils down to playing with spark settings.
Now consider the first block: I think this is a classic case of merge sort
or a reduce tree. You already tried to improve this by submitting jobs in
parallel using executor pool/Callable etc.
To further improve the parallelization, I suggest you use a reduce tree
like approach. For example, lets say you want to compute sum of all
elements of an array in parallel. The way its solved for a GPU like
platform is you divide your input array initially in chunks of 2, compute
those n/2 sums parallely on separate threads and save the result in the
first of the two elements. In the next iteration, you compute n/4 sums
parallely of the earlier sums and so on till you are left with only two
elements whose sum gives you final sum.
You are performing many sequential unionAll operations for inputs.size()
avro files. Assuming the unionAll() on DataFrame is blocking (and not a
simple transformation like on RDDs) and actually performs the union
operation, you will certainly benefit by parallelizing this loop. You may
change the loop to something like below:
// pseudo code only
int n = inputs.size()
// initialize executor
executor = new FixedThreadPoolExecutor(n/2)
dfInput = new DataFrame[n/2]
for(int i =0;i < n/2;i++) {
executor.submit(new Runnable() {
public void run() {
// union of i and i+n/2
// showing [] only to bring out array access. Replace with
dfInput(i) in your code
dfInput[i] = sqlContext.load(inputPaths.get(i),
"com.databricks.spark.avro").unionAll(sqlContext.load(inputsPath.get(i +
n/2), "com.databricks.spark.avro"))
}
});
}
executor.awaitTermination(0, TimeUnit.SECONDS)
int steps = log(n)/log(2.0)
for(s = 2; s < steps;s++) {
int stride = n/(1 << s); // n/(2^s)
for(int i = 0;i < stride;i++) {
executor.submit(new Runnable() {
public void run() {
// union of i and i+n/2
// showing [] only to bring out array access. Replace with
dfInput(i) and dfInput(i+stride) in your code
dfInput[i] = dfInput[i].unionAll(dfInput[i + stride])
}
});
}
executor.awaitTermination(0, TimeUnit.SECONDS)
}
Let me know if it helped.
-Kiran
On Thu, Jun 4, 2015 at 8:57 PM, James Aley wrote:
> Thanks for the confirmation! We're quite new to Spark, so a little
> reassurance is a good thing to have sometimes :-)
>
> The thing that's concerning me at the moment is that my job doesn't seem
> to run any faster with more compute resources added to the cluster, and
> this is proving a little tricky to debug. There are a lot of variables, so
> here's what we've tried already and the apparent impact. If anyone has any
> further suggestions, we'd love to hear!
>
> * Increase the "minimum" number of output files (targetPartitions above),
> so that input groups smaller than our minimum chunk size can still be
> worked on by more than one executor. This does measurably speed things up,
> but obviously it's a trade-off, as the original goal for this job is to
> merge our data into fewer, larger files.
>
> * Submit many jobs in parallel, by running the above code in a Callable,
> on an executor pool. This seems to help, to some extent, but I'm not sure
> what else needs to be configured alongside it -- driver threads, scheduling
> policy, etc. We set scheduling to "FAIR" when doing this, as that seemed
> like the right approach, but we're not 100% confident. It seemed to help
> quite substantially anyway, so perhaps this just needs further tuning?
>
> * Increasing executors, RAM, etc. This doesn't make a difference by itself
> for this job, so I'm thinking we're already not fully utilising the
> resources we have in a smaller cluster.
>
> Again, any recommendations appreciated. Thanks for the help!
>
>
> James.
>
> On 4 June 2015 at 15:00, Eugen Cepoi wrote:
>
>> Hi
>>
>> 2015-06-04 15:29 GMT+02:00 James Aley :
>>
>>> Hi,
>>>
>>> We have a load of Avro data coming into our data systems in the form of
>>> relatively small files, which we're merging into larger Parquet files with
>>> Spark. I've been following the docs and the approach I'm taking seemed
>>> fairly obvious, and pleasingly simple, but I'm wondering if perhaps it's
>>> not the most optimal approach.
>>>
>>> I was wondering if anyone on this list might have some advice to make to
>>> make this job as efficient as possible. Here's some code:
>>>
>>> DataFrame dfInput = sqlContext.load(inputPaths.get(0),
>>> "com.databricks.spark.avro");
>>> long totalSize
Re: Optimisation advice for Avro->Parquet merge job
Thanks for the confirmation! We're quite new to Spark, so a little
reassurance is a good thing to have sometimes :-)
The thing that's concerning me at the moment is that my job doesn't seem to
run any faster with more compute resources added to the cluster, and this
is proving a little tricky to debug. There are a lot of variables, so
here's what we've tried already and the apparent impact. If anyone has any
further suggestions, we'd love to hear!
* Increase the "minimum" number of output files (targetPartitions above),
so that input groups smaller than our minimum chunk size can still be
worked on by more than one executor. This does measurably speed things up,
but obviously it's a trade-off, as the original goal for this job is to
merge our data into fewer, larger files.
* Submit many jobs in parallel, by running the above code in a Callable, on
an executor pool. This seems to help, to some extent, but I'm not sure what
else needs to be configured alongside it -- driver threads, scheduling
policy, etc. We set scheduling to "FAIR" when doing this, as that seemed
like the right approach, but we're not 100% confident. It seemed to help
quite substantially anyway, so perhaps this just needs further tuning?
* Increasing executors, RAM, etc. This doesn't make a difference by itself
for this job, so I'm thinking we're already not fully utilising the
resources we have in a smaller cluster.
Again, any recommendations appreciated. Thanks for the help!
James.
On 4 June 2015 at 15:00, Eugen Cepoi wrote:
> Hi
>
> 2015-06-04 15:29 GMT+02:00 James Aley :
>
>> Hi,
>>
>> We have a load of Avro data coming into our data systems in the form of
>> relatively small files, which we're merging into larger Parquet files with
>> Spark. I've been following the docs and the approach I'm taking seemed
>> fairly obvious, and pleasingly simple, but I'm wondering if perhaps it's
>> not the most optimal approach.
>>
>> I was wondering if anyone on this list might have some advice to make to
>> make this job as efficient as possible. Here's some code:
>>
>> DataFrame dfInput = sqlContext.load(inputPaths.get(0),
>> "com.databricks.spark.avro");
>> long totalSize = getDirSize(inputPaths.get(0));
>>
>> for (int i = 1; i < inputs.size(); ++i) {
>> dfInput = dfInput.unionAll(sqlContext.load(inputPaths.get(i),
>> "com.databricks.spark.avro"));
>> totalSize += getDirSize(inputPaths.get(i));
>> }
>>
>> int targetPartitions = (int) Math.max(2L, totalSize / TARGET_SIZE_BYTES);
>> DataFrame outputFrame;
>>
>> // Note: HADOOP-10456 impacts us, as we're stuck on 2.4.0 in EMR, hence
>> // the synchronize block below. Suggestions welcome here too! :-)
>> synchronized (this) {
>> RDD inputRdd = dfInput.rdd().coalesce(targetPartitions, false,
>> null);
>> outputFrame = sqlContext.createDataFrame(inputRdd, dfInput.schema());
>> }
>>
>> outputFrame.save(outputPath, "parquet", SaveMode.Overwrite);
>>
>> Here are some things bothering me:
>>
>>- Conversion to an RDD and back again so that we can use coalesce()
>>to reduce the number of partitions. This is because we read that
>>repartition() is not as efficient as coalesce(), and local micro
>> benchmarks
>>seemed to somewhat confirm that this was faster. Is this really a good
>> idea
>>though? Should we be doing something else?
>>
>> Repartition uses coalesce but with a forced shuffle step. Its just a
> shortcut for coalesce(xxx, true)
> Doing a coalesce sounds correct, I'd do the same :) Note that if you add
> the shuffle step, then your partitions should be better balanced.
>
>>
>>- Usage of unionAll() - this is the only way I could find to join the
>>separate data sets into a single data frame to save as Parquet. Is there a
>>better way?
>>
>> When using directly the inputformats you can do this
> FileInputFormat.addInputPath, it should perform at least as good as union.
>
>>
>>- Do I need to be using the DataFrame API at all? I'm not querying
>>any data here, so the nice API for SQL-like transformations of the data
>>isn't being used. The DataFrame API just seemed like the path of least
>>resistance for working with Avro and Parquet. Would there be any advantage
>>to using hadoopRDD() with the appropriate Input/Output formats?
>>
>>
>>
> Using directly the input/outputformats sounds viable. But the snippet you
> show seems clean enough and I am not sure there would be much value in
> making something (maybe) slightly faster but harder to understand.
>
>
> Eugen
>
> Any advice or tips greatly appreciated!
>>
>>
>> James.
>>
>>
>>
>
Re: Optimisation advice for Avro->Parquet merge job
Hi
2015-06-04 15:29 GMT+02:00 James Aley :
> Hi,
>
> We have a load of Avro data coming into our data systems in the form of
> relatively small files, which we're merging into larger Parquet files with
> Spark. I've been following the docs and the approach I'm taking seemed
> fairly obvious, and pleasingly simple, but I'm wondering if perhaps it's
> not the most optimal approach.
>
> I was wondering if anyone on this list might have some advice to make to
> make this job as efficient as possible. Here's some code:
>
> DataFrame dfInput = sqlContext.load(inputPaths.get(0),
> "com.databricks.spark.avro");
> long totalSize = getDirSize(inputPaths.get(0));
>
> for (int i = 1; i < inputs.size(); ++i) {
> dfInput = dfInput.unionAll(sqlContext.load(inputPaths.get(i),
> "com.databricks.spark.avro"));
> totalSize += getDirSize(inputPaths.get(i));
> }
>
> int targetPartitions = (int) Math.max(2L, totalSize / TARGET_SIZE_BYTES);
> DataFrame outputFrame;
>
> // Note: HADOOP-10456 impacts us, as we're stuck on 2.4.0 in EMR, hence
> // the synchronize block below. Suggestions welcome here too! :-)
> synchronized (this) {
> RDD inputRdd = dfInput.rdd().coalesce(targetPartitions, false,
> null);
> outputFrame = sqlContext.createDataFrame(inputRdd, dfInput.schema());
> }
>
> outputFrame.save(outputPath, "parquet", SaveMode.Overwrite);
>
> Here are some things bothering me:
>
>- Conversion to an RDD and back again so that we can use coalesce() to
>reduce the number of partitions. This is because we read that repartition()
>is not as efficient as coalesce(), and local micro benchmarks seemed to
>somewhat confirm that this was faster. Is this really a good idea though?
>Should we be doing something else?
>
> Repartition uses coalesce but with a forced shuffle step. Its just a
shortcut for coalesce(xxx, true)
Doing a coalesce sounds correct, I'd do the same :) Note that if you add
the shuffle step, then your partitions should be better balanced.
>
>- Usage of unionAll() - this is the only way I could find to join the
>separate data sets into a single data frame to save as Parquet. Is there a
>better way?
>
> When using directly the inputformats you can do this
FileInputFormat.addInputPath, it should perform at least as good as union.
>
>- Do I need to be using the DataFrame API at all? I'm not querying any
>data here, so the nice API for SQL-like transformations of the data isn't
>being used. The DataFrame API just seemed like the path of least resistance
>for working with Avro and Parquet. Would there be any advantage to using
>hadoopRDD() with the appropriate Input/Output formats?
>
>
>
Using directly the input/outputformats sounds viable. But the snippet you
show seems clean enough and I am not sure there would be much value in
making something (maybe) slightly faster but harder to understand.
Eugen
Any advice or tips greatly appreciated!
>
>
> James.
>
>
>
Optimisation advice for Avro->Parquet merge job
Hi,
We have a load of Avro data coming into our data systems in the form of
relatively small files, which we're merging into larger Parquet files with
Spark. I've been following the docs and the approach I'm taking seemed
fairly obvious, and pleasingly simple, but I'm wondering if perhaps it's
not the most optimal approach.
I was wondering if anyone on this list might have some advice to make to
make this job as efficient as possible. Here's some code:
DataFrame dfInput = sqlContext.load(inputPaths.get(0),
"com.databricks.spark.avro");
long totalSize = getDirSize(inputPaths.get(0));
for (int i = 1; i < inputs.size(); ++i) {
dfInput = dfInput.unionAll(sqlContext.load(inputPaths.get(i),
"com.databricks.spark.avro"));
totalSize += getDirSize(inputPaths.get(i));
}
int targetPartitions = (int) Math.max(2L, totalSize / TARGET_SIZE_BYTES);
DataFrame outputFrame;
// Note: HADOOP-10456 impacts us, as we're stuck on 2.4.0 in EMR, hence
// the synchronize block below. Suggestions welcome here too! :-)
synchronized (this) {
RDD inputRdd = dfInput.rdd().coalesce(targetPartitions, false,
null);
outputFrame = sqlContext.createDataFrame(inputRdd, dfInput.schema());
}
outputFrame.save(outputPath, "parquet", SaveMode.Overwrite);
Here are some things bothering me:
- Conversion to an RDD and back again so that we can use coalesce() to
reduce the number of partitions. This is because we read that repartition()
is not as efficient as coalesce(), and local micro benchmarks seemed to
somewhat confirm that this was faster. Is this really a good idea though?
Should we be doing something else?
- Usage of unionAll() - this is the only way I could find to join the
separate data sets into a single data frame to save as Parquet. Is there a
better way?
- Do I need to be using the DataFrame API at all? I'm not querying any
data here, so the nice API for SQL-like transformations of the data isn't
being used. The DataFrame API just seemed like the path of least resistance
for working with Avro and Parquet. Would there be any advantage to using
hadoopRDD() with the appropriate Input/Output formats?
Any advice or tips greatly appreciated!
James.
