5 Benefits of Hybrid HPC for HE
From processing terabytes of research data to running the latest in 3D rendering software, higher education institutions have some of the most diverse and demanding computing requirements. Whether storing, processing or analysing, the data needs across courses and departments require a dedicated resource that can handle complex tasks. Many larger universities already have dedicated on-site high-performance computing (HPC) facilities, where they can offer a range of services to individuals, groups and organisations that need serious computational power.
For smaller institutions, it might not be feasible to offer this kind of resource, and even for larger ones, dedicated on-premise HPC can be difficult to scale up or even down as requirements and demand shifts.
Moving to a hybrid HPC and infrastructure is a major trend in both the public and private sectors at the moment. Hybrid basically means that you operate some services on-premise, and some in the cloud. There are a number of benefits to doing things in this way - here are just 5.
The first major benefit of hybrid HPC is one we’ve already alluded to. With a hybrid infrastructure, as you’re harnessing the power of both your on-premise devices as well as hosting some resources in the cloud, it’s easily scalable as your requirements grow. Scaling a traditional HPC environment can be expensive, and may require a significant amount of additional space to house it, capital investment to procure it, as well as a resource to monitor and maintain it. With a hybrid infrastructure, you can scale your cloud requirements on-demand, increasing or decreasing storage as necessary.
Using an Infrastructure-as-a-Service (IaaS) model, cloud-based storage can offer you increased security for your data. This is because your provider will be responsible for back-up, disaster recovery and protection against cyber attacks, rather than this falling on the institution to manage on-premise. This physical separation from the university campus and critical data also mean that in the event of on-premise downtime or an incident, your data is available and completely secure while you get back up and running. You can also back up on-premise data to the cloud at a frequency that suits your institution and your users, easily recoverable at any time.
A disadvantage of a traditional HPC setup may be that the user or users will need to be on-premise in order to access the resources. Shifting to a hybrid HPC infrastructure means being able to offer certain services anytime, anywhere via the cloud. This would be particularly useful for distance learning students looking to store data. It could even be useful for collaboration - users all over the world can have access to the same resources allowing them to seamlessly work on research and projects together without even having to be in the same country. It also means it can be managed from anywhere too - your network managers and administrators, who might be based or working on a different campus, don’t need to physically get themselves to the server room or data centre to fix an issue or do some proactive work - they can login from anywhere and do what they need to do.
The world of HPC moves very quickly. Even if your institution’s data centre was running the latest processors, storage and networking when you invested in your hardware, this can be superseded by new technology very quickly. Keeping up with the rate of change with a purely on-premise environment simply isn’t feasible - it would require a significant investment of time, money and resource. Using a hybrid model can give you more ready access to the latest server technology powered by best-in-class components, so you always know that you’re operating at the very bleeding edge of what’s possible.
Shifting some of your processes and services to the cloud is more efficient in a number of ways. It helps you save on space, power consumption, and as a result, budget. As you don’t need as much physical on-premise hardware, you cut down on maintenance costs. A hybrid infrastructure is more efficient from a performance perspective too, giving you faster delivery, better application performance and faster cryptography for securing data.In 2017, Centerprise secured a £1.8m contract with CERN, assembling, testing and delivering over 125 Intel energy-efficient rack-mountable systems at the headquarters in Switzerland to support physics data processing and control systems. We understand HPC at the highest level. To learn more about how we could help your data centre to perform as well as one of the world’s most prolific research organisations,Get in Touch