HP Powers Linux Clusters


Showing a convergence between open source and mainstream enterprise
technology, HP unveiled a new file system that uses the
company’s hardware and Linux to deliver up to 100 times more bandwidth than
traditional clusters.


The HP StorageWorks Scalable File Share (HP SFS), which includes the
company’s ProLiant servers and StorageWorks disk arrays, allows bandwidth to
be shared by distributing files in parallel across clusters of servers and
storage devices. The system was designed to bust bandwidth constraints
in high-performance computing environments.


Simon Towers, a technical director in the office of the CTO at HP, said the system is based on HP’s
“storage grid” architecture, allowing applications to see a single file
system image regardless of the number of servers or storage devices
connected to it.


The system also protects against hardware failures with redundant hardware
and built-in fail-over and recovery. SFS can span dozens to thousands of
clustered Linux servers, making it ideal to run distributed applications for
science and engineering projects.


Towers told internetnews.com products like SFS are important to HP as a storage vendor because of the drastic increase in the amount of unstructured data — e-mail, video clips and PDFs — has grown. File systems are needed to manage this data.


“In a way, HP has sort of leveraged all the open source programmers out there to rearchitect how file systems are built and lay that on top of storage grid architecture,” Towers said.


Companies such as IBM and HP, along with software makers
like Red Hat and SuSE, have been working to broaden the
sphere of the Linux operating system in the commercial enterprise, as well
as scale the software out for HPC projects.


The companies want to push the envelope and prove that Linux
can no longer be banished to the back rooms of research labs. However,
they recognize Linux as a viable alternative to find the crack in
Windows’ armor.


That the SFS is powered by Linux is a statement from HP that
Linux is viable for large-scale computing. SFS is the first commercial
product to use Lustre, a new Linux clustering technology developed by HP,
the U.S. Department of Energy (DoE) and Cluster File Systems.


The Lustre protocol used in SFS already powers some of the world’s largest
HPC environments, including the DoE Pacific Northwest National Laboratory
(PNNL), where it eliminates bandwidth bottlenecks and saves users hours of
time copying files across distributed file systems.

One of the 10 largest Linux clusters in the world according to the Top500
supercomputing rankings, PNNL’s HP Linux super
cluster, clocks in at more than 11 teraflops (one trillion floating point
operations per second) and sustains more than 3.2 gigabytes per second of
bandwidth running production loads on a single 53-terabyte, Lustre-based file
share.


IBM makes a
file system in its storage product line, called TotalStorage SAN File System,
that runs Red Hat Linux Enterprise Server 3.0 and Sun Solaris 9. HP’s
SFS is based on StorageWorks grid architecture, which allows storage
services to be delivered across a centrally managed system.


The SFS news comes just two days after the Top500 project posted
its latest list of leading supercomputers. HP was second to IBM in total
share of systems to make the cut, grabbing a number two ranking with 28
percent of all systems.

News Around the Web