M7 Dramatically Boosts Apache HBase™ Application Performance Across Every Workload

Contributed by

2 min read

As part of the latest MapR M7 release for NoSQL and Apache Hadoop, MapR conducted benchmark tests to measure and validate its Apache HBase™ application performance. The MapR M7 Edition dramatically boosted the performance of applications originally written to run with HBase with upwards of 10x better throughput while eliminating latency spikes.

The benchmark tests were performed using Yahoo Cloud Serving Benchmark (YCSB) - the widely accepted standard for NoSQL performance testing. A copy of the full benchmark report is available here for those interested in the details. The results that measured both throughput and latency showed the following:


  • HBase applications running on M7 performed better than those running on other distributions across all of the workloads including put, get, range scan and mixed fetch-update workloads.
  • M7 performance is far more pronounced for random row retrieval scenarios showcasing M7's sturdier database design, with 10x better performance.
  • For mixed workloads with prebuilt caches, M7 performance is more than twice that of HBase applications on other distributions.


M7 Low Latency Advantage Graph
Read Latency: Lower is Better. YCSB Mixed Workload (50% Update-50%Read); 10 Nodes; 2TB (1K row size); 10 second moving average; Y-axis cap = 400 msec

This figure provides a graphical representation of the latency differences with M7. Not only is the latency significantly lower with M7, what is perhaps more important is the level of consistency. One of the biggest issues impacting online applications using HBase is the lack of consistent latency. Large latency swings are hard to plan around and can significantly impact usability of applications. With M7, not only does the low latency translate into more effective online applications, but also the consistency makes it much easier to deploy applications successfully.

To get started with M7, click here and see the results for yourself.

This blog post was published September 25, 2013.

50,000+ of the smartest have already joined!

Stay ahead of the bleeding edge...get the best of Big Data in your inbox.

Get our latest posts in your inbox

Subscribe Now