kudu-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benjamin Kim <b...@amobee.com>
Subject Re: Performance Question
Date Mon, 11 Jul 2016 17:35:29 GMT
Over the weekend, a tablet server went down. It’s not coming back up. So, I decommissioned
it and removed it from the cluster. Then, I restarted Kudu because I was getting a timeout
 exception trying to do counts on the table. Now, when I try again. I get the same error.

16/07/11 17:32:36 WARN scheduler.TaskSetManager: Lost task 468.3 in stage 0.0 (TID 603, prod-dc1-datanode167.pdc1i.gradientx.com<http://prod-dc1-datanode167.pdc1i.gradientx.com>):
com.stumbleupon.async.TimeoutException: Timed out after 30000ms when joining Deferred@712342716(state=PAUSED,
result=Deferred@1765902299, callback=passthrough -> scanner opened -> wakeup thread
Executor task launch worker-2, errback=openScanner errback -> passthrough -> wakeup
thread Executor task launch worker-2)
at com.stumbleupon.async.Deferred.doJoin(Deferred.java:1177)
at com.stumbleupon.async.Deferred.join(Deferred.java:1045)
at org.kududb.client.KuduScanner.nextRows(KuduScanner.java:57)
at org.kududb.spark.kudu.RowResultIteratorScala.hasNext(KuduRDD.scala:99)
at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:327)
at org.apache.spark.sql.execution.aggregate.TungstenAggregate$$anonfun$doExecute$1$$anonfun$2.apply(TungstenAggregate.scala:88)
at org.apache.spark.sql.execution.aggregate.TungstenAggregate$$anonfun$doExecute$1$$anonfun$2.apply(TungstenAggregate.scala:86)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitions$1$$anonfun$apply$20.apply(RDD.scala:710)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitions$1$$anonfun$apply$20.apply(RDD.scala:710)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:73)
at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:41)
at org.apache.spark.scheduler.Task.run(Task.scala:89)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:214)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)

Does anyone know how to recover from this?

Benjamin Kim
Data Solutions Architect

[a•mo•bee] (n.) the company defining digital marketing.

Mobile: +1 818 635 2900
3250 Ocean Park Blvd, Suite 200  |  Santa Monica, CA 90405  |  www.amobee.com<http://www.amobee.com/>

On Jul 6, 2016, at 9:46 AM, Dan Burkert <dan@cloudera.com<mailto:dan@cloudera.com>>

On Wed, Jul 6, 2016 at 7:05 AM, Benjamin Kim <bbuild11@gmail.com<mailto:bbuild11@gmail.com>>
Over the weekend, the row count is up to <500M. I will give it another few days to get
to 1B rows. I still get consistent times ~15s for doing row counts despite the amount of data

On another note, I got a solicitation email from SnappyData to evaluate their product. They
claim to be the “Spark Data Store” with tight integration with Spark executors. It claims
to be an OLTP and OLAP system with being an in-memory data store first then to disk. After
going to several Spark events, it would seem that this is the new “hot” area for vendors.
They all (MemSQL, Redis, Aerospike, Datastax, etc.) claim to be the best "Spark Data Store”.
I’m wondering if Kudu will become this too? With the performance I’ve seen so far, it
would seem that it can be a contender. All that is needed is a hardened Spark connector package,
I would think. The next evaluation I will be conducting is to see if SnappyData’s claims
are valid by doing my own tests.

It's hard to compare Kudu against any other data store without a lot of analysis and thorough
benchmarking, but it is certainly a goal of Kudu to be a great platform for ingesting and
analyzing data through Spark.  Up till this point most of the Spark work has been community
driven, but more thorough integration testing of the Spark connector is going to be a focus
going forward.

- Dan


On Jun 15, 2016, at 12:47 AM, Todd Lipcon <todd@cloudera.com<mailto:todd@cloudera.com>>

Hi Benjamin,

What workload are you using for benchmarks? Using spark or something more custom? rdd or data
frame or SQL, etc? Maybe you can share the schema and some queries



On Jun 15, 2016 8:10 AM, "Benjamin Kim" <bbuild11@gmail.com<mailto:bbuild11@gmail.com>>
Hi Todd,

Now that Kudu 0.9.0 is out. I have done some tests. Already, I am impressed. Compared to HBase,
read and write performance are better. Write performance has the greatest improvement (>
4x), while read is > 1.5x. Albeit, these are only preliminary tests. Do you know of a way
to really do some conclusive tests? I want to see if I can match your results on my 50 node


On May 30, 2016, at 10:33 AM, Todd Lipcon <todd@cloudera.com<mailto:todd@cloudera.com>>

On Sat, May 28, 2016 at 7:12 AM, Benjamin Kim <bbuild11@gmail.com<mailto:bbuild11@gmail.com>>

It sounds like Kudu can possibly top or match those numbers put out by Aerospike. Do you have
any performance statistics published or any instructions as to measure them myself as good
way to test? In addition, this will be a test using Spark, so should I wait for Kudu version
0.9.0 where support will be built in?

We don't have a lot of benchmarks published yet, especially on the write side. I've found
that thorough cross-system benchmarks are very difficult to do fairly and accurately, and
often times users end up misguided if they pay too much attention to them :) So, given a finite
number of developers working on Kudu, I think we've tended to spend more time on the project
itself and less time focusing on "competition". I'm sure there are use cases where Kudu will
beat out Aerospike, and probably use cases where Aerospike will beat Kudu as well.

From my perspective, it would be great if you can share some details of your workload, especially
if there are some areas you're finding Kudu lacking. Maybe we can spot some easy code changes
we could make to improve performance, or suggest a tuning variable you could change.


On May 27, 2016, at 9:19 PM, Todd Lipcon <todd@cloudera.com<mailto:todd@cloudera.com>>

On Fri, May 27, 2016 at 8:20 PM, Benjamin Kim <bbuild11@gmail.com<mailto:bbuild11@gmail.com>>
Hi Mike,

First of all, thanks for the link. It looks like an interesting read. I checked that Aerospike
is currently at version, and in the article, they are evaluating version 3.5.4. The
main thing that impressed me was their claim that they can beat Cassandra and HBase by 8x
for writing and 25x for reading. Their big claim to fame is that Aerospike can write 1M records
per second with only 50 nodes. I wanted to see if this is real.

1M records per second on 50 nodes is pretty doable by Kudu as well, depending on the size
of your records and the insertion order. I've been playing with a ~70 node cluster recently
and seen 1M+ writes/second sustained, and bursting above 4M. These are 1KB rows with 11 columns,
and with pretty old HDD-only nodes. I think newer flash-based nodes could do better.

To answer your questions, we have a DMP with user profiles with many attributes. We create
segmentation information off of these attributes to classify them. Then, we can target advertising
appropriately for our sales department. Much of the data processing is for applying models
on all or if not most of every profile’s attributes to find similarities (nearest neighbor/clustering)
over a large number of rows when batch processing or a small subset of rows for quick online
scoring. So, our use case is a typical advanced analytics scenario. We have tried HBase, but
it doesn’t work well for these types of analytics.

I read, that Aerospike in the release notes, they did do many improvements for batch and scan

I wonder what your thoughts are for using Kudu for this.

Sounds like a good Kudu use case to me. I've heard great things about Aerospike for the low
latency random access portion, but I've also heard that it's _very_ expensive, and not particularly
suited to the columnar scan workload. Lastly, I think the Apache license of Kudu is much more
appealing than the AGPL3 used by Aerospike. But, that's not really a direct answer to the
performance question :)


On May 27, 2016, at 6:21 PM, Mike Percy <mpercy@cloudera.com<mailto:mpercy@cloudera.com>>

Have you considered whether you have a scan heavy or a random access heavy workload? Have
you considered whether you always access / update a whole row vs only a partial row? Kudu
is a column store so has some awesome performance characteristics when you are doing a lot
of scanning of just a couple of columns.

I don't know the answer to your question but if your concern is performance then I would be
interested in seeing comparisons from a perf perspective on certain workloads.

Finally, a year ago Aerospike did quite poorly in a Jepsen test: https://aphyr.com/posts/324-jepsen-aerospike

I wonder if they have addressed any of those issues.


On Friday, May 27, 2016, Benjamin Kim <bbuild11@gmail.com<mailto:bbuild11@gmail.com>>
I am just curious. How will Kudu compare with Aerospike (http://www.aerospike.com<http://www.aerospike.com/>)?
I went to a Spark Roadshow and found out about this piece of software. It appears to fit our
use case perfectly since we are an ad-tech company trying to leverage our user profiles data.
Plus, it already has a Spark connector and has a SQL-like client. The tables can be accessed
using Spark SQL DataFrames and, also, made into SQL tables for direct use with Spark SQL ODBC/JDBC
Thriftserver. I see from the work done here http://gerrit.cloudera.org:8080/#/c/2992/ that
the Spark integration is well underway and, from the looks of it lately, almost complete.
I would prefer to use Kudu since we are already a Cloudera shop, and Kudu is easy to deploy
and configure using Cloudera Manager. I also hope that some of Aerospike’s speed optimization
techniques can make it into Kudu in the future, if they have not been already thought of or

Just some thoughts…


Mike Percy
Software Engineer, Cloudera

Todd Lipcon
Software Engineer, Cloudera

Todd Lipcon
Software Engineer, Cloudera

View raw message