Whamcloud flogs wild Lustre pig into obedience with data whip
It's the only language these open brutes understand
ISC 2012 Whamcloud is adding enterprise-type data management features to the open-source Lustre parallel file system.
Whamcloud is a Lustre development company counting some Lustre inventors in its ranks. Lustre is a parallel file system enabling hundreds and thousands of supercomputer compute nodes to access file data at the same time, avoiding the bottlenecking associated with a single filer head as is the case with an NFS filesystem.
As open-source software, Lustre competes with proprietary parallel file systems such as PanFS from Panasas and IBM's GPFS. But it's generally recognised that Lustre is harder to deploy and manage than either PanFS or GPFS.
Chroma Enterprise is now out of beta test and available to partners. It is integrated with Lustre and provides:
- automated device discovery and menu-driven configuration to simplify Lustre configuration, creation and setup;
- easy-to-use browser-based monitoring and management; and
- flexibility to choose storage arrays that suit the workload.
Robert Read, Chroma engineering manager at Whamcloud, said: "One of the biggest problems with Lustre has always been managing it. Answering ‘How is my file system today?’ should be easy. There are a lot of moving parts in a Lustre file system and checking them all requires time, patience, and expertise. We built Chroma Enterprise to address these exact issues."
We're told that there's a lot more of this sort of development coming from Whamcloud.
Whamcloud said it has been encouraged by the DDN and NetApp announcements that they're shipping Chroma-based products. It wants Lustre usage to expand into enterprise markets as the big data trend brings HPC computing into the commercial mainstream.
This is a first step and GPFS and PanFS will be thinking there's lipstick there but Lustre is still a pig underneath. Whamcloud will no doubt prefer the idea of Lustre as an ugly duckling being transformed into an enterprise swan.
There's more information here. ®
Sponsored: Hyper-scale data management