With ISC’12 coming up in June, insideHPC is taking a closer look at European HPC vendors. This week we caught up with Dr.Tony Kent from ClusterVision to discuss how the company is leveraging it’s core strengths in high performance computing to deliver something called the Total Cluster Lifecycle.
insideHPC: How did ClusterVision get started?
Dr.Tony Kent: ClusterVision was founded approximately 10 years ago by Matthijs van Leeuwen, now CEO of Bright Computing, and Alex Ninaber, now CTO of ClusterVision. Both had a background in computational science and HPC applications, and formed ClusterVision to address the growing market need for a professional partner who had niche expertise and strong connections within the HPC industry. Still based in Amsterdam, The Netherlands, but now with offices throughout Europe, ClusterVision has since grown into a multi-million international operation, and has designed and built some of the fastest and most complex computational and database clusters in Europe, including many Top500 systems. ClusterVision is now led by Alex Ninaber CTO, Christopher Huggins Sales & Marketing Director, and Frank van der Hout Financial Director. Matthijs van Leeuwen left ClusterVision to form Bright Computing, the developers of the cluster management suite Bright Cluster Manager, which is now included by ClusterVision as one of the critical components in our overall cluster solution.
insideHPC: Is HPC your core business?
Dr.Tony Kent: It’s more than just our core business – it’s our daily life. At ClusterVision we are 100% dedicated to HPCC (High Performance Compute Cluster) solutions. All of our customer facing staff come from technical computing or HPC application backgrounds, and we believe that our customers benefit from the undivided focus and attention to detail which such a dedication brings. Having said that, the cluster itself is not our motivation. At ClusterVision we have a saying, “It’s not the size of the cluster, it’s the size of the science” – it’s the HPC applications which our systems enable where our customer satisfaction really lies.
insideHPC: You have an impressive customer list including a number of universities in Europe. Have you also gained traction in the industrial sector as well?
Dr.Tony Kent: Yes, our early engagements focussed on the academic sector, – the early adopters who were driving the HPC technologies and applications. Universities and other research establishments are still an important market focus for ClusterVision – systems technologies advance rapidly and compute intensive applications and data requirements are becoming increasingly demanding, so we work closely with both new and existing customers in academia to ensure that their HPCC facilities keep pace. However, today ClusterVision also serves a wide range of commercial markets, from Government Laboratories, to Manufacturing, Healthcare, Exploration, Energy, Media and Finance. While the HPC sector is still seeing low double-digit growth, some of the newer commercial sectors are growing substantially faster. As we look today, some of our largest and most exciting projects are now with companies involved with applications in oil and gas exploration, automotive and aerospace engineering, and financial market analysis.
insideHPC: Your web site talks about the “total cluster lifecycle.” What do you mean by that?
Dr.Tony Kent: Total Cluster Lifecycle refers to the full design, build, and management timeline of cluster ownership. We realise that for our customers the purchase and delivery of a cluster system is just the start of their process. Customers are increasingly aware of the total ownership implications of their systems, both in terms of the operational costs such as power consumption, but also for less tangible on-going considerations like sustainability, process change requirements and skills training. Our first action is always to gain a deep understanding of our customers’ situation – not just their processing or data requirements, but their legacy systems and investments, current capabilities, applications, objectives, and of course their constraints, economically, technologically, and operationally. In this way we are able to design a solution with an optimal balance between investment and performance, but also one where informed decisions can be made upfront about the long-term ownership. Once delivered, the degree of operational management requirement varies, so we can provide anything from initial provisioning to full scale remote cluster administration. The HPCC market moves quickly, so future proofing current systems for emerging developments, and being on-hand to guide customers through upscale and enhancement projects is also an important part of our offering.
insideHPC: As a European vendor, how does ClusterVision’s offering differ from the other solutions out there in the marketplace?
Dr.Tony Kent: Being European in itself is one differentiator. Although we have customers all over the world, our core business and customer focus is in Europe, where we feel we are best able to offer an agile and personal level of service. Perhaps more important is our consideration of a complete HPCC solution – systems, software and services. High performing cluster solutions require the harmonious integration of a complex collection of hardware and software components, and the expert services which ensure that these are selected, assembled, and managed in an optimal way. Where others may focus on individual aspects of this process, we are able to offer our customers a complete, one-stop, turn-key solution. Our strong partnerships with all of the leading brand manufacturers means that our systems are never compromised in terms of component quality, while our independence ensures that we are ideally placed to provide our customers with impartial advice on technology performance and selection, as is best suited to their needs.
insideHPC: You offer support services. How critical is support to your HPC customers?
Dr.Tony Kent: Yes, support is an essential ingredient of the total cluster lifecycle concept. Most people think of support in terms of post-delivery hotline services, and repair and maintenance arrangements, and of course these are part of our standard delivery. Customers can even select different levels of support service to reflect the varying degrees of criticality of each component to their operation. However, our dedication to total cluster solutions means that we can also offer a wide range of other value-add services, throughout the cluster lifecycle. Services such as future proofed design configurations, industry standard certification, application benchmarking and optimisation, provisioning and remote administration. We also offer training to both systems administrators and end users to transfer the skills they will need to operate and utilise their cluster to its maximum capability. In addition to providing a rationalised single point of support contact, all of these services are primarily designed to ease the overall burden of cluster ownership, allowing our customers to focus on what they do best – their scientific applications.
Alexey Komkov: We strive to maintain the stability of financial flows while diversifying our profit sources. The last financial year saw increased investment in the development of new technologies, products and services. In 2010-2011, many potential customers delayed contracts for delivery of HPC-systems because of the global financial crisis, and therefore we expect significant growth in sales in 2012.
I recently saw some estimates that show we should hit exascale supercomputer performance by around 2018. That seems a bit ambitious – if not stunningly optimistic – and the search to get some perspective led me on an hours-long meander through supercomputing history, plus what I like to call “Fun With Spreadsheets.”
