Originally presented at: https://spark-summit.org/eu-2017/events/the-state-of-apache-spark-in-the-cloud/ Cloud providers currently offer convenient on-demand managed big data clusters (PaaS) with a pay-as-you-go model. In PaaS, analytical engines such as Spark and Hive come ready to use, with a general-purpose configuration and upgrade management. Over the last year, the Spark framework and APIs have been evolving very rapidly, with major improvements on performance and the release of v2, making it challenging to keep up-to-date production services both on-premises and in the cloud for compatibility and stability. Nicolas Poggi evaluates the out-of-the-box support for Spark and compares the offerings, reliability, scalability, and price-performance from major PaaS providers, including Azure HDinsight, Amazon Web Services EMR, Google Dataproc with an on-premises commodity cluster as baseline. Nicolas uses BigBench, the brand new standard (TPCx-BB) for big data systems, with both Spark and Hive implementations for benchmarking the systems. BigBench combines SQL queries, MapReduce, user code (UDF), and machine learning, which makes it ideal to stress Spark libraries (SparkSQL, DataFrames, MLlib, etc.). The work is framed within the ALOJA research project, which features an open source benchmarking and analysis platform that has been recently extended to support SQL-on-Hadoop engines and BigBench. The ALOJA project aims to lower the total cost of ownership (TCO) of big data deployments and study their performance characteristics for optimization. Nicolas highlights how to easily repeat the benchmarks through ALOJA and benefit from BigBench to optimize your Spark cluster for advanced users. The work is a continuation of a paper to be published at the IEEE Big Data 16 conference.

1. The state of in the cloud
Nicolas Poggi
Oct 2017
____ __
/ __/__ ___ _____/ /__
_ / _ / _ `/ __/ '_/
/___/ .__/_,_/_/ /_/_
/_/

2. Outline
1. Intro
1. PaaS Cloud
2. BigBench
2. Part I – Scalability
1. Hive vs. Spark
3. Part II – Additional experiments
1. Versions, Concurrency, 10TB
4. Summary
2

3. Motivation
• 2016 SQL-on-Hadoop paper and presentations
• Focused on Hive, due to SparkSQL not being ready to use in PaaS
• Used benchmark (TPC-H)
• Early 2017, BigBench testing Hive and Spark and TPC-TC paper
• New code available in May for MLlib2 compatibility
• Goals:
Evaluate the current out-of-the-box experience of Spark v2 in PaaS cloud
• Using Hive as baseline
3

4. Platform-as-a-Service Spark
• Simplified management
• Cloud-based managed Hadoop services
• Ready to use Spark, Hive, …
• Deploys in minutes, on-demand, elastic
• Pay-as-you-go pricing model
• Decoupled compute and storage
• Optimized for general purpose
• Fined tuned to the cloud provider architecture
4

5. Surveyed PaaS services
• Amazon Elastic Map Reduce (EMR)
• Released: Apr 2009
• OS: Amazon Linux AMI (RHEL-like)
• Spark 2.1.0 and Hive 2.1 (Tez)
• VM: M4.2xlarge (32GB RAM)
• Google Cloud DataProc (GCD)
• Released: Feb 2016
• OS: Debian 8.4
• Spark 2.1.0 (preview), Hive 2.1 (M/R)
• VM: n1-standard-8 (30GB RAM)
• Azure HDInsight (HDI)
• Released: Oct 2013
• OS: Ubuntu 16.04 (HDP-based)
• Spark 2.1.0 and 1.6.3, Hive 1.2 (Tez)
• VM: D4v2 (28GB RAM)
• Target deployment 128-cores:
• 16 data nodes with 8-cores each
• Master node with 16-cores
• Decoupled storage only
• EBS, WASB, GCS
5

6. What is BigBench (TPCx-BB)
• End-to-end application level benchmark specification
• result of many years of collaboration of industry and academia
• Covers most Big Data Analytical properties (3Vs)
• 30 business use cases for a retailer company
• Merchandizing,
• pricing,
• customers …
• Defines data scale factors
• 1GB to PBs
6
Retailer database

7. Sequential Hive vs Spark 2.1
Queries 1-30 on Spark 2.1 (power runs)
Query 1 Query 2 …. Query 30
Welcome to
____ __
/ __/__ ___ _____/ /__
_ / _ / _ `/ __/ '_/
/___/ .__/_,_/_/ /_/_ version 2.1.0
/_/

8. BB 1GB-1TB Scalability Dataproc:
Hive 2.1 (M/R) vs Spark 2.1

9. BB 1GB-1TB Scalability EMR:
Hive 2.1 (Tez) vs Spark 2.1

10. BB 1GB-1TB Scalability HDI:
Hive 1.2 (Tez) vs Spark 2.1

11. BB 1GB-1TB Scalability: Hive vs Spark 2.1
All providers

12. BB 1TB Power runs : Hive vs Spark 2.1 (ALL)

13. CPU % Q5 (ML) in Hive and Spark (HDI)
13
• Hive (MLlib2) • Spark (MLlib2)
Time (s) Time (s) - 2X faster

14. Radar charts – query characterization
• Useful for displaying multivariate
data (5 resources)
• Quickly identify similarities and
differences.
• From example
• Hive and Spark
• Only Disk Write is similar
• Hive consumes more MEM and CPU
• Spark read more from disk (DISK_R)
• And moderately more network
Sample radar chart for Q7 in EMR at 1TB 14

15. BB 1TB Query 5 (ML) providers comparison
Hive (MLlib2) Spark (MLlib2)
15

16. Other comparisons:
10TB SQL-Only
2.0.2 vs 2.1.0
1.6.3 vs 2.1.0
MLlib v1 vs v2
16

17. BB 1GB-10TB Scalability SQL-only queries
Hive
Spark

18. BigBench 1GB-1TB: Spark 2.0.2 vs 2.1.0 (CDP)
Notes:
Spark 2.1 a bit faster at
small scales, slower at
100 GB and 1 TB on the
UDF/NLP queries
2.1 faster up
to 100GB
Slower at 1TB

19. BigBench 1GB-1TB: Spark 1.6.3 vs 2.1.0
MLlib 1 vs 2.1 MLlib 2(HDI)
Notes:
• Spark 2.1 is always
faster than 1.6.3 in
HDI
• MLlilb 2 using
dataframes over RDDs
is only slightly faster
than V1.

20. Concurrency runs (throughput)
SQL-only: 100GB – 1TB 512-core cluster
2020

21. BB Throughput at 100GB 8 streams SQL-only
(512-cores)

22. BB Throughput at 1TB 4 streams SQL-only
(512-cores)

23. Conclusions
• All providers have up to date (2.1.0) and well tuned versions of Spark
• They could run BigBench up to 1TB on medium-sized cluster,
• [Almost] Out-of-the box
• Performance similar among providers for similar cluster types and disk configs
• Difference according to scale (and pricing)
• Spark 2.1.0 is faster than previous versions
• Also MLlib 2 with dataframes
• But improvements within the 30% range
• Hive (+Tez + MLlib) are still slightly faster than Spark at lower scales for sequential
• But Spark significantly faster at high data scales and concurrency
• BigBench has been useful to stress a cluster with different workloads
• Highlights config problems fast and stresses scale limits
• Helpful for tuning the clusters
23

24. Resources and references
BigBench and ALOJA
• BigBench Spark 2 branch (thanks Christoph
and Michael from bankmark.de):
• https://github.com/carabolic/Big-Data-
Benchmark-for-Big-Bench/tree/spark2
• Original BigBench Implementation
repository
• https://github.com/intel-hadoop/Big-Data-
Benchmark-for-Big-Bench
• ALOJA benchmarking platform
• https://github.com/Aloja/aloja
• ALOJA fork of BigBench (adds support for
HDI and fixes spark)
• https://github.com/Aloja/Big-Data-Benchmark-
for-Big-Bench
Papers and slides
• https://www.slideshare.net/ni_po
• Characterizing TPCx-BB Queries,
Hive, and Spark in Multi-Cloud
Environments – N. Poggi et. Al
• TPC-TC 2017
• The State of SQL-on-Hadoop in the
Cloud – N. Poggi et. al.
• IEEE Big Data 2016
• https://doi.org/10.1109/BigData.2016
.7840751
24

25. Thanks, questions?
Follow up / feedback : Npoggi@ac.upc.edu
Twitter: ni_po
The state of in the cloud
____ __
/ __/__ ___ _____/ /__
_ / _ / _ `/ __/ '_/
/___/ .__/_,_/_/ /_/_
/_/

26. Extra slides
26

27. BB 1TB Query 2 (M/R) providers comparison
Hive Spark
27

28. Spark config
EMR CDP HDI
Java version OpenJDK 1.8.0_121 OpenJDK 1.8.0_121 OpenJDK 1.8.0_131
Spark version 2.1.0 2.1 2.1.0.2.6.0.2-76
Driver memory 5G 5G 5G
Executor memory 5G 10G 4G
Executor cores 4 4 3
Executor instances Dynamic Dynamic 20
dynamicAllocation
enabled
TRUE TRUE FALSE
Executor
memoryOverhead
Default (384MB) 1,117 MB Default (384MB)
28