Nicolas Poggi

Over the years, there has been extensive efforts to improve Apache Spark SQL performance. This talk will introduce the new Adaptive Query Execution (AQE) framework and how it can automatically improve user query performance. AQE leverages query runtime statistics to dynamically guide Spark's execution as queries run along. The talk will go over the main features in AQE and provide examples on how it can improve on the previous static query plans. Finally, we'll present the significant improvements we have seen on the TPC-DS benchmark with AQE.

Examples of the new runtime optimizations include selecting the right join type (broadcast-hash-join vs. sort-merge-join), dealing with data skew, and automatically selecting the number of shuffle (reducer) partitions.

Non-Volatile-Memory express (NVMe) standard promises and order of magnitude faster storage than regular SSDs, while at the same time being more economical than regular RAM on TB/$. This talk evaluates the use cases and benefits of NVMe drives for its use in Big Data clusters with HBase and Hadoop HDFS.

First, we benchmark the different drives using system level tools (FIO) to get maximum expected values for each different device type and set expectations. Second, we explore the different options and use cases of HBase storage and benchmark the different setups. And finally, we evaluate the speedups obtained by the NVMe technology for the different Big Data use cases from the YCSB benchmark.

In summary, while the NVMe drives show up to 8x speedup in best case scenarios, testing the cost-efficiency of new device technologies is not straightforward in Big Data, where we need to overcome system level caching to measure the maximum benefits.

BigBench is the brand new standard for benchmarking and testing Big Data systems. This talk first introduces BigBench and how problems can it solve. Then, presents both Hive and Spark benchmark results with with their respective 1 and 2 versions under different configurations. Results are further classified by use cases, showing where each platform shines (or doesn't), and why, based on performance metrics and log-file analysis.

BigBench is the brand new standard (TPCx-BB) for benchmarking and testing Big Data systems. The BigBench specification describes several application use cases combining the need for SQL queries, Map/Reduce, user code (UDF), Machine Learning, and even streaming. From the available implementation, we can test the different framework combinations such as Hadoop+Hive+Tez (with Mahout) and Spark (SparkSQL+SparkML) in their different versions and configurations, helping us to spot problems and possible optimizations of our data stacks.

This talk first introduces BigBench and how problems can it solve. Then, presents both Hive and Spark benchmark results with with their respective 1 and 2 versions under different configurations including: Tez, LLAP, and file formats. Experiments are run on Cloud and On-Prem clusters of different numbers of nodes and testing data scales, taking into account interactive and batch usage. Results are further classified by use cases, showing where each platform shines (or doesn't), and why, based on performance metrics and log-file analysis. The talk concludes with the main findings, the scalability and limits of each framework.

This talks will present how to perform benchmarking of Big Data systems, from low-powered devices, HPC clusters, to cloud IaaS and PaaS. It will guide participants on how to define clusters, select benchmark suites and configuration with the ALOJA open source tools. ALOJA (http://aloja.bsc.es) is a research initiative from the Barcelona Supercomputing Center to explore new cost-effective hardware architectures for Big Data. During its first year, ALOJA's benchmarking efforts have produced the largest public repository with over 50,000 Hadoop benchmark runs. The searchable repository features different applications for Hadoop, software configurations, data sizes, and more than 100 different hardware deployment options.

The talk will cover the following topics:

-Cluster definition and automated deployment on local (vagrant clusters), on-prem and cloud environments

-Automating and orchestrating OS, Hadoop, JVM configuration across clusters

-Benchmark selection and iteration of configurations

-Metrics collections, results gathering, and importing

-Advanced data views for aggregate results with filters

-Analyzing performance results and selecting configurations