Performance Evaluation of Cloudera Impala GA
5 likes3,544 views
This document evaluates the performance of Cloudera Impala 1.0 using a modified HiBench benchmark on a 11-node cluster. It finds that Parquet files compressed with Snappy provide the fastest query performance, completing in 16.2 seconds on average. The order of joined tables can also impact performance. Impala shows significant performance improvements over Hive, with the fastest query being over 14 times faster. Additional functions like UDFs and window functions are recommended for future Impala extensions.
![Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p /
Benchmark Result (Hive)
cited from “Performance evaluation of Cloudera impala 0.6 beta...”
8
0 50 100 150 200 250
No Comp.
Gzip
Snappy
Gzip
Snappy
TextFileSequenceFileRCFile
235.843
227.883
213.616
234.289
197.894
Avg. Job Latency [sec]](https://image.slidesharecdn.com/performanceevaluationofclouderaimpalaga-130501054452-phpapp02/85/Performance-Evaluation-of-Cloudera-Impala-GA-8-320.jpg)
![Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p /
Benchmark Result (Impala)
9
0 50 100 150 200 250
No Comp.
Gzip
Snappy
Gzip
Snappy
Snappy
Text
File
Sequence
FileRCFile
Parquet
File
36.61
29.736
24.024
26.083
19.586
16.2
Avg. Job Latency [sec]](https://image.slidesharecdn.com/performanceevaluationofclouderaimpalaga-130501054452-phpapp02/85/Performance-Evaluation-of-Cloudera-Impala-GA-9-320.jpg)
Performance Evaluation of Cloudera Impala GA
- 1. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / 1 1 Performance Evaluation of Cloudera impala 1.0 May 1, 2013 CELLANT Corp. R&D Strategy Division Yukinori SUDA @sudabon
- 2. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / v Support for a subset of ANSI-‐‑‒92 SQL v CREATE, ALTER, SELECT, INSERT, JOIN, and subqueries v Support for partitioned joins, fully distributed aggregations, and fully distributed top-‐‑‒n queries v Support for a variety of data formats: v Hadoop native (Apache Avro, SequenceFile, RCFile with Snappy, GZIP, BZIP, or uncompressed) v text (uncompressed or LZO-‐‑‒compressed) v Parquet (Snappy or uncompressed) v Support for all CDH4 64-‐‑‒bit packages: v RHEL 6.2/5.7, Ubuntu, Debian, SLES v Connectivity via JDBC, ODBC, Hue GUI, or command-‐‑‒line shell v Kerberos authentication and MR/Impala resource isolation v etc Cloudera Impala GA was released !! 2
- 3. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / Our System Environment 3 v Install using Cloudera Manager Free Edition 4.5.2 Master Slave 11 Servers All servers are connected with 1Gbps Ethernet through an L2 switch Active NameNode DataNode TaskTracker Impalad Stand-‐‑‒by NameNode JobTracker statestored 3 Servers DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad DataNode TaskTracker Impalad
- 4. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / v CPU l Intel Core 2 Duo 2.13 GHz with Hyper Threading v Memory l 4GB v Disk l 7,200 rpm SATA mechanical Hard Disk Drive * 1 v OS l Cent OS 6.2 Our “wimpy” Server Specification 4
- 5. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / v Use CDH4.2.1 + Impala version 1.0 v Use hivebench in open-‐‑‒sourced benchmark tool “HiBench” l https://github.com/hibench v Modified datasets to 1/10 scale l Default configuration generates table with 1 billion rows v Modified query sentence l Deleted “INSERT INTO TABLE …” to evaluate read-‐‑‒only performance v Combines a few storage format with a few compression method l TextFile, SequenceFile, RCFile, ParquestFile l No compression, Gzip, Snappy v Comparison with job query latency v Average job latency over 5 measurements Benchmark 5
- 6. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / • Uservisits table – 100 million rows – 16,895 MB as TextFile – Table Definitions • sourceIP string • destURL string • visitDate string • adRevenue double • userAgent string • countryCode string • languageCode string • searchWord string • duration int • Rankings table – 12 million rows – 744 MB as TextFile – Table Definitions • pageURL string • pageRank int • avgDuration int Modified Datasets 6
- 7. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / SELECT sourceIP, sum(adRevenue) as totalRevenue, avg(pageRank) FROM rankings_̲t R JOIN ( SELECT sourceIP, destURL, adRevenue FROM uservisits_̲t UV WHERE (datediff(UV.visitDate, '1999-‐‑‒01-‐‑‒01')>=0 AND datediff(UV.visitDate, '2000-‐‑‒01-‐‑‒01')<=0) ) NUV ON (R.pageURL = NUV.destURL) group by sourceIP order by totalRevenue DESC limit 1; Modified Query 7
- 8. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / Benchmark Result (Hive) cited from “Performance evaluation of Cloudera impala 0.6 beta...” 8 0 50 100 150 200 250 No Comp. Gzip Snappy Gzip Snappy TextFileSequenceFileRCFile 235.843 227.883 213.616 234.289 197.894 Avg. Job Latency [sec]
- 9. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / Benchmark Result (Impala) 9 0 50 100 150 200 250 No Comp. Gzip Snappy Gzip Snappy Snappy Text File Sequence FileRCFile Parquet File 36.61 29.736 24.024 26.083 19.586 16.2 Avg. Job Latency [sec]
- 10. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / v Exchange the order of JOINed Tables like below SELECT sourceIP, sum(adRevenue) as totalRevenue, avg(pageRank) FROM (SELECT sourceIP, destURL, adRevenue FROM uservisits_̲ps UV WHERE (datediff(UV.visitDate, '1999-‐‑‒01-‐‑‒01')>=0 AND datediff(UV.visitDate, '2000-‐‑‒01-‐‑‒01')<=0)) NUV JOIN rankings_̲ps R ON (R.pageURL = NUV.destURL) group by sourceIP order by totalRevenue DESC limit 1; v Result l Parquet compressed as Snappy: 34.374 sec Additional Experiments 10
- 11. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / v Parquet + Snappy is the fastest v Specifically, l ParquetFile compressed as Snappy: 16.2 sec v Need to take care the order of JOINed tables v Hope for future extension l Support UDF l Window Function l etc Conclusion 11
- 12. Copyright © CELLANT Corp. All Rights Reserved. h t t p : / / w w w . c e l l a n t . j p / 12 Letʼ’s try it out on your envrionment!! Thanks!