What are you looking for ?
Infinidat
Articles_top

MapR Accelerates HBase Applications

Quadrupling throughput and eliminating latency spikes

Leveraging its Hadoop performance, MapR Technologies, Inc. has updated its M7 edition to improve HBase application performance with throughput that is 4-10x faster while eliminating latency spikes.

                              M7 Low latency advantage
MapR,Turbocharges,HBase Applications
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 = 400ms

HBase applications can now benefit from MapR’s high performance platform to address one of the major issues for on-line applications, consistent read latencies in the less than 20ms range across varying workloads.

Our customers are moving Hadoop from pilot adoption and project use to mainstream enterprise deployments,” said John Schroeder, CEO and cofounder, MapR. “MapR customers are experiencing the same reliability and enterprise performance with our distribution as they have seen with the Oracle platform at a fraction of the cost.”

Companies have been migrating to M7 from Oracle, MySQL and other NoSQL databases. These companies are choosing MapR to provide the reliability and performance characteristics that they need for mission critical and online applications.

MapR provides consistent reliability that is essential to our business and takes HBase performance and manageability to a whole new level,” said Scott Russmann, Solutionary, Inc.‘s director of software engineering. “With MapR, our systems can survive failures, all while running in production with no down time. MapR absolutely broke out as an innovator in the Hadoop world.”

MapR is able to achieve such performance through:

  • Architecture that persists table structure at the file-system layer;
  • No compactions (I/O storms) for HBase applications;
  • Workload-aware splits for HBase applications;
  • Direct writes to disk (vs. writing to an external file system);
  • Disk and network compression; and
  • C++ implementation that does not suffer from garbage collection problems seen with Java applications.

HBase’s biggest limitation is regarding consistent and deterministic performance,” said John Webster, senior partner, Evaluator Group. “MapR has addressed these limitations and provides support for online applications in a production data center setting.”

MapR M7 Edition is available for download for on-premise use as well as for cloud deployments with Amazon EMR and the Google Cloud Platform.

Articles_bottom
AIC
ATTO
OPEN-E