HPC: multiple vendors key to successful partnering

People joining hands as a symbol of teamwork

It’s no secret that a few years ago our primary vendor partner was IBM. It was the right thing to do at the time. ‘Our bird in hand’ was steady, proven and valuable to us and our customers. IBM provided us with world leading hardware and software to design and build robust HPC systems for our customers (they still do in many cases). We also spent less time recruiting and managing partners and more time serving customers. It was then, and still is, a very successful partnership.

However, times change. After 2008, the world went mad, the pound went from $2 to $1.3 and these unwanted macro economic conditions were causing a lot of issues for our customers. We needed to help our customers in the best way possible, so we went looking for ‘birds in the bush’ and broadened our partner base from only being with IBM to having additional partners for each of the components within our solutions.

It was a significant move away – the polar opposite in fact - from our previous strategy, but the market moves and it became evident that being so heavily aligned to a single vendor partner for a period of time restricted choices for us and our customers. The subsequent acquisition of IBM’s x86 server technology by Lenovo certainly proves that nothing stands still for long and we need to be flexible and agile in our attitude.

As time has moved on, we have developed new partnerships. In the server market, we have deployed HPC systems with Lenovo at The Cloud Infrastructure for Microbial Bioinformatics (CLIMB) project and University of Oxford. However, we also do a lot of work with Fujitsu and completed a successful deployment at The Wellcome Trust Centre for Human Genetics [WTCHG].

In storage, we work with DDN, NetApp, Fujitsu, Lenovo and IBM. The storage solutions in a lot of cases are more differentiated according to capability because storage is a more varied product than servers. Nonetheless, we have designed and deployed systems at WTCHG and Public Health England using DDN technology, as another example.

The WTCHG based at the University of Oxford uses a Fujitsu high-performance BX900 blade-based cluster with Mellanox InfiniBand and DDN storage system. It was designed to support the genetics research of 25 groups and more than 100 researchers. Dr Robert Esnouf, Head of Research Computing at the Wellcome Trust said, “Research is driving our adoption of HPC. Compared to this time last year, our researchers can put through five times more work and are doing so on a machine with the same energy footprint. With the support of OCF and its hardware partners like Fujitsu and IBM, we’re now fully armed to meet today’s challenges.”

As we move further into 2016, I believe public sector and commercial organisations will want HPC systems built using best of breed solutions. There will be no single solution right for everyone. Our experience of working in the channel is one that needs flexibility and agility – our customers want the latest and greatest HPC systems, regardless of vendor, so sticking with one vendor doesn’t always make sense. But, of course, building lasting partnerships is important too – it’s a fine balance between working closely with key vendors and finding and recruiting new ones to make sure you remain at the top of the innovation tree.

Julian Fielden is managing director of OCF