Platform Computing's OCS Extends Clustering with Open Source Tools
Published: October 3, 2006
by Timothy Prickett Morgan
As one of the pioneers in grid computing for supercomputing workloads, Platform Computing is well acquainted with open source software. While the company's flagship LSF workload management software remains the touchstone for grids, it has not gone open source with its products. But with the Open Cluster Stack, Platform is getting closer.
Platform OCS is being peddled on Linux platforms in conjunction with server partner Dell, but there is no reason it cannot run on other X86 and X64 iron. OCS can run on top of Red Hat Enterprise Linux or the CentOS variant of Red Hat. OCS is a bunch of software that runs on this, which includes modules for development tools and utilities, file systems, system management, workload management, and resource management--both at the server node and cluster levels.
Platform has been spearheading the commercialization of an open source cluster management project called Rocks, which was created by the San Diego Supercomputing Center, which is owned and administered by the University of California at San Diego. Platform has also created an open source cluster workload management tool called Lava, which is central to the OCS product and which provides some of the functionality of LSF. LSF is still arguably a more sophisticated cluster workload manager, and it can plug into the OCS framework, as can popular open source tools such as Ganglia (which monitors workloads on nodes and clusters), Clumon (which is a system admin dashboard for monitoring grids), MatTool (for managing disks and DNS servers), and about a dozen other snap-ins, including the IBRIX parallel file system and Intel's compilers. Eventually, the Nagios system management tool will plug into the OCS framework, too.
OSC is, according to Gary Tyreman, vice president of the open cluster group at Platform, meant to be a competitor to the Grid Engine product from Sun Microsystems and the OpenPBS product from Altair Engineering. And at $150 per server node, OCS is priced much more aggressively than LSF, which probably means that the product will see much greater uptake in the market. LSF costs $399 per CPU, and on two-socket or four-socket server nodes, that can add up pretty fast in an HPC cluster.
While OCS is only available on Red Hat right now, Tyreman says that Platform is evaluating ports to Novell's SUSE Linux Enterprise Server, Ubuntu LTS, and Red Flag Linux DC Server. And if Sun keeps getting traction with its Solaris 10/OpenSolaris platform, Platform might even consider a port to Solaris as well.
Platform Gets EAL Security Certification for Grid Products
Server Makers Push Linux As Linux Pulls Them