Using BigBench to compare Hive and Spark (short version)
0 likes419 views
The document discusses BigBench, an open-source benchmark standard for big data that evaluates Hive and Spark using various queries from TPC-DS and additional machine learning and NLP use cases. It highlights the performance improvements of Hive on Tez compared to MapReduce and Spark, and suggests that Apache Tez combined with Spark MLlib is the best approach for production. The benchmark tests a 100 GB dataset across different configurations, revealing Hive on Tez as the most efficient option.
![What is BigBench (TPCx-BB1)?
• Specification-based benchmark with an open-source implementation, proposed as
the first Big Data benchmark standard.
• BigBench covers all major Big Data characteristics
• Volume -> Scale factor.
• Velocity -> Table refresh.
• Variety -> Data type disparity.
• Extension of TPC-DS.
• Borrows 10 pure QL queries from TCP-DS.
• Added BigData tables and use cases: Machine Learning, Natural Language Processing, ...
• Can support multiple implementations, multiple BigData engines and table formats.
• Can execute multiple parallel streams.
• Defines scale factors for data.
• Tested: 100 GB.
2[1]: http://www.tpc.org/tpc_documents_current_versions/pdf/tpcx-bb_v1.2.0.pdf](https://image.slidesharecdn.com/usingbigbenchtocomparehiveandspark-170205120522/85/Using-BigBench-to-compare-Hive-and-Spark-short-version-2-320.jpg)
Using BigBench to compare Hive and Spark (short version)
- 1. Using BigBench to compare Hive and Spark Alejandro Montero, Nicolas Poggi 1
- 2. What is BigBench (TPCx-BB1)? • Specification-based benchmark with an open-source implementation, proposed as the first Big Data benchmark standard. • BigBench covers all major Big Data characteristics • Volume -> Scale factor. • Velocity -> Table refresh. • Variety -> Data type disparity. • Extension of TPC-DS. • Borrows 10 pure QL queries from TCP-DS. • Added BigData tables and use cases: Machine Learning, Natural Language Processing, ... • Can support multiple implementations, multiple BigData engines and table formats. • Can execute multiple parallel streams. • Defines scale factors for data. • Tested: 100 GB. 2[1]: http://www.tpc.org/tpc_documents_current_versions/pdf/tpcx-bb_v1.2.0.pdf
- 3. BigBench – Overview 3 Unstructured DataStructured Data Semi-Structured Data Marketprice Items Sales Web Page Customers Reviews Web Log Workload: • 14 Pure QL queries. • 10 Borrowed from TPC-DS. • 4 Queries with MapReduce pre-processing. • 7 Natural Language Processing Queries. • 5 Machine Learning Queries.
- 4. BigBench v1.2 – Reference Implementation HDFS Hive Metastore MapReduce Tez Spark Yarn Hive Spark SQL Mahout ML Custom Spark MLlibApplication SQL Engine Table Metastore Execution Engine Filesystem Benchmarked systems: • Hive + MapReduce + Mahout • Hive + MapReduce + Spark_MLlib • Hive + Tez + Mahout • Hive + Tez + Spark_MLlib • Spark SQL + Mahout • Spark SQL + Spark_MLlib • Spark 2 SQL + Mahout Work in progress: • Hive 2 • Spark 2 SQL + Spark_MLlib
- 5. The cluster – HDInsight PaaS 5 Model HDInsight D4v3 # Head nodes 2 # Working nodes 4 # Zookeeper nodes 3 CPU Intel(R) Xeon(R) CPU E5-2673 v3 8 x 2,4 GHz cores RAM 28 GB HDFS Remote Software HortonWorks Data Platform 2.5 Spark config 1 executor/working node 3 cores/executor
- 6. Pure QL 6 6223 1601 1848 1457 0 1000 2000 3000 4000 5000 6000 7000 Hive_MR Hive_tez Spark_1.6.2 Spark_2.0.1 Timeinseconds Average of three executions using 100 GB Scale Factor
- 7. Query 12 CPU behavior 7 Tez Spark 1.6.2 Spark 2.0.2 Average of three executions using 100 GB Scale Factor
- 8. Custom Reducers 8 2815 1122 1629 1466 0 500 1000 1500 2000 2500 3000 Hive_MR Hive_tez Spark_1.6.2 Spark_2.0.1 Timeinseconds Average of three executions using 100 GB Scale Factor
- 9. Natural Language Processing 9 5004 1100 2289 1913 0 1000 2000 3000 4000 5000 6000 Hive_MR Hive_Tez Spark_1.6.2 Spark_2.0.1 Timeinseconds Average of three executions using 100 GB Scale Factor
- 10. Machine Learning 10 3550 1613 3769 1898 4045 1937 3390 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Hive_MR + Mahout Hive_MR + Spark_ml Hive_tez + Mahout Hive_Tez + Spark_ML Spark_1.6.2 + Mahout Spark_1.6.2 + Spark_ML Spark_2.0.1 + Mahout Timeinseconds Average of three executions using 100 GB Scale Factor
- 11. 11 Aggregated Results 6223 6223 1601 1623 1848 1951 1457 2815 2815 1122 1137 1629 1489 1466 5004 5004 1100 1082 2289 2356 1913 3550 1613 3769 1898 4045 1937 3390 17592 15655 7592 5740 9811 7733 8226 0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 Hive_MR + Mahout Hive_MR + Spark_ML Hive_tez + Mahout Hive_Tez + Spark_ML Spark_1.6.2 + Mahout Spark_1.6.2 + Spark_ML Spark_2.0.1 + Mahout Timeinseconds Pure-QL Custom Reducers NLP ML Total Average of three executions using 100 GB Scale Factor
- 12. Conclusions • Hive on Tez greatly improves SQL performance over Hive on MapReduce. • It is also faster than Hive on spark 1. • Hive on spark 2 is slightly faster. • The Spark implementation is based on hive… • Spark MLlib has an excellent performance over Mahout. • Best production combination: Apache Tez for SQL + Spark MLlib for Machine Learning. 12
- 13. Thanks, questions? Follow up / feedback : Alejandro.montero@bsc.es Using BigBench to compare Hive and Spark 13