HPC

Transforming Xyratex

The data storage market has always proven an exciting market to be a part of. Less than ten years ago the industry was dealing with new mobile drives, doubling capacity numbers, emergence of new mobile devices (Smartphones, etc) and significant growth rates in the Enterprise. Before that it was the constant evolution of interface technologies, several performance types (recall the introduction of 15k disk drives?) and the focus on Enterprise quality.

Introducing ClusterStor™ 6000

Tired of your compute cluster waiting on file system I/O? Are you interested in seeing high performance data storage that will make your HPC environment scream? Well today, we just announced our latest edition to the ClusterStor series of products and it is the ultimate in HPC data storage. ClusterStor 6000 can deliver huge performance at massive scale in a lot less space, power and cooling than any other storage system. So what?…well despite their claims most other HPC storage systems are all bark and no bite.

Webinar - InfiniBand-based Storage for Efficient and Scalable High Performance Computing

I had the recent opportunity on January 25,2012 to participate with our valuable partner Mellanox Technologies, Ltd. in a joint educational webinar. Mellanox is a leading supplier of high-performance, end-to-end connectivity solutions for data center servers and storage systems and Xyratex currently utilizes their InfiniBand technology as part of our ClusterStor™ 3000 storage solutions.

Storage Performance: Block vs. File System

File system performance is often a major component of overall HPC system performance, and is heavily dependent on the nature of the application(s) generating the load. To achieve optimal performance, the underlying file system configuration must be balanced to match the application characteristics. Hence the basis for Lustre. However, regardless of the style of file system (such as UFS, NTFS, ZFS, or any of the parallel file systems used today), the underlying disk structure dictates the performance capabilities of these file systems.

Why Lustre?

Lustre is not the only scalable parallel file system in the HPC world but, during it’s 10 year lifespan, it still dominates in eight out of the top ten HPC systems in the world and is in more than 70% of the Top500.  Lustre is open source so it is not tied to a single vendor and has strong support in the community. One of the main reasons for Lustre’s dominance is its inherent scaling and the impressive roadmap for future performance scaling. 

Heading to SBAC-PAD-2001 in Brazil . . .

This Thursday, I’ll be speaking at South America’s largest HPC conference about breaking the I/O bottleneck in HPC storage so please join us if you’re planning on to be at the conference. I feel that as an industry, we face many architectural challenges around creating a compute/storage architecture that can perform and scale to meet the requirements of tomorrow’s large Petascale and eventually, Exascale systems.

Benefits of Parallel File Systems for HPC Performance

When architecting HPC storage solutions, there are really two options to look at:  a network file system (NFS) solution or a parallel file system. 

Live From the HPC User Forum

Hello from San Diego! This week looks to be a busy one for the HPC community. With so much happening in this industry, it’s hard to stay up to date with all of the recent installations, new technologies and industry initiatives. That’s why I like participating in events like IDC’s HPC User Forum.

Yearning for Simpler Times

Why does storage for HPC have to be so hard? Building HPC data storage using traditional servers, storage systems and network fabrics (SAN) gets in the way of simplicity, performance and resiliency for those that just care about getting to their research results.