by Brian Beeler

STEC, Mellanox, AMAX and Zettaset Deliver PHAT-Data40G High Performance Hadoop Appliance

A four-horsemen of partners including STEC, Mellanox, AMAX and Zettaset are demonstrating a new appliance designed to deliver record-breaking performance for Hadoop users. The PHAT-Data40G as it's known, is powered by four STEC 980GB PCIe storage accelerators, Mellanox's 40GbE interconnect solutions, AMAX's Intel Xeon E-based servers and Zettaset's installation, management, and security tools.

STEC MLC PCIe SSA Specifications

  • Capacity: 480GB/980GB
  • Type: X4 PCIe 2.0
  • Form Factor: HH-HL
  • Sustained Read Throughput: 1.4GB/s
  • Sustained Write Throughput: 600MB/s
  • Read (512B): 44,000 IOPS
  • Write (512B): 163,000 IOPS
  • Operational Temperature: 0º to 55ºC (Commercial)

The PHAT-Data40G is a simple to deploy turnkey solution that provides Hadoop users with the combination of top-end performance with unparalleled ease of use. Compared to other PCIe storage-based solutions, the TeraSort benchmark shows the PHAT-Data40G is able to sort a 1TB dataset in under 27 minutes with 5 machines. These results are 20% faster than previous best-in-class performance, with half the PCIe storage acceleration cost.

 

The PHAT-Data40G incorporates Zettaset's installation, management, and security tools, providing customers with a unified solution for Apache Hadoop acceleration in less than 12U of rack space. The solution offers compatibility with Java-based MapReduce applications and supports heterogeneous environments and open standards

AMAX PHAT-Data appliances include 2 name nodes configured with DRBD/Heartbeat or RSF-1 in master-slave configuration with a dedicated heartbeat network. The auxiliary NameNode automatically takes over in case of a failure and the MapReduce engine offers automatic job recovery without the need to restart in case of JobTracker/TaskTracker failure.

PHAT-Data40G Key Features

  • Turnkey solution with supporting software stack
  • Supports file systems and data sources including HDFS, GPFS and any file system with HDFS plugin
  • Self-monitoring and troubleshooting capabilities
  • Support for up to 300 job trackers across a shared pool of resources
  • Application and job level predictability via sophisticated job scheduler to guarantee SLA requirements
  • Support for homogenous and heterogeneous network environments
  • 10G Ethernet
  • Hypervisor support coming soon

Availability
The PHAT-Data40G is available now and is distributed through AMAX Technologies. 

Product Page

Discuss This Story