Boost Your HPC Power with Hybrid Cloud Bursting

Share This

When it comes to high-performance computing resources, IT management faces two major problems: capability and capacity.

Capability: Sometimes you need to access new technology to provide leading-edge tools to your users to meet project requirements. Common scenarios would include testing new hardware technology or accessing exotic capabilities you don’t need frequently, like a multi-GPU compute system.

Capacity: In a typical production environment, demand varies with time. Variations in demand can be driven by a range of factors, and on-premises HPC resources with fixed capacity may not fit an organization’s precise usage requirements. Sometimes workload will exceed on-premises capacity, while at other times there may be excess capacity that goes unused. Purchasing additional hardware has long lead times, and it’s useful to have strategies that allow us to access additional capability until we can meet those requirements.

Solving the Problems of Capacity and Capability

I recently hosted a webinar, now available on demand, about hybrid cloud bursting as a solution for augmenting your HPC power.

Hybrid cloud bursting can help your organization meet fluctuating resource requirements and manage excess workload demand. It can help you meet project deadlines, get results on time, and keep customer commitments — all part of making the right decisions for your business. It can also allow you to access new technologies quickly to assess their impact on your business.

Hybrid cloud bursting also lets us think about meeting resource requirements in a different way, by leveraging the cloud more extensively. It’s one of the tradeoffs we, as customers, make when making decisions about the balance between investments in on-premises capabilities vs. those in the cloud.

Cloud Bursting Terminology

Here are a few definitions related to cloud bursting as a solution for changing organizational HPC resource requirements:

  • Private cloud – Your own HPC resources and data centre.
  • Public cloud – Someone else’s data centre and hardware. Altair partners with major cloud vendors that have strong security models.
  • Hybrid cloud bursting – Extend your private cloud into the public cloud in order to increase capacity and/or capability while maintaining a strong security model and control over usage and costs.

Tools for Hybrid Cloud Bursting

We can help you get started with cloud bursting.  The Altair PBS Works™ suite toolkit provides all the software you need for end-to-end workload and resource management — from user access to administrative control and workload optimization.

  • Altair Control™ – Burst to one or more clouds, configure access to your public cloud account, and create multiple bursting scenarios for control and flexibility.
  • Altair Access™ – Submit hybrid cloud workloads in an easy, drag-and-drop interface via web, desktop, or mobile.
  • Altair PBS Professional™ – Intelligently route your HPC workloads and control cloud access.

We use Altair’s knowledge and expertise to help customers migrate their software stacks to the cloud and run them efficiently.

In addition to choosing a platform and tools, it’s important to select the right workload for bursting to the cloud. An effective workload to burst may be one with small input datasets and small output datasets that we can route to the cloud. Bursting criteria may also relate to the license model of the tools being used, or it may have to do with scalability; for example, if we lack the ability to run large jobs that utilize multiple nodes on-premises, we can route them to the cloud to access those capabilities. 

Solution Architecture and Configuration

You can bring your own public cloud account or use an Altair account if you’re not ready to manage your own, or if you simply don’t want to. We’ll build your setup for one or more clouds and make sure your network is configured optimally.

With Altair’s hybrid cloud bursting solution architecture, you can choose classical hybrid bursting that allows HPC workloads to burst to the cloud when resources require it and policies allow, or a custom appliance that gives you an end-to-end solution with no need for your own private cloud data centre.

Hybrid cloud bursting allows rapid and flexible resource growth, and it’s a transparent solution for users that provides a managed end-to-end data lifecycle. Watch our on-demand webinar to see a cloud bursting demo and learn more about hybrid cloud bursting, including:

  • How to configure hybrid cloud bursting
  • How to submit cloud workloads
  • How to decide on cloud workloads
  • Public cloud bursting controls
  • Licensing models

Connect with an Altair expert to learn more about hybrid cloud bursting.

Ian Littlewood

Ian Littlewood is a long-time Altairian of nearly 20 years. He holds a degree in mechanical engineering and began work in engineering design on industrial and automotive machinery in 1991, then transitioned to CAE analysis — and HPC systems — while working primarily on car suspension components, cross car beams, engine mounts, and other projects for key global OEMs. This started a long history of using workload management tools and tuning application sets on HPC systems.
At Altair, Ian has supported Altair HyperWorks™, where knowledge of desktop and HPC application tuning is critical, and managed Altair’s UK HPC footprint. He moved into workload management with Altair PBS Professional™ 15 years ago to support, implement, and architect customer solutions for major customers. After a period in sales, he transitioned to product management to bring his breadth of knowledge to bear in mapping customer requirements into the software development process of Altair PBS Works Suite.
Ian Littlewood

Latest posts by Ian Littlewood (see all)


Share This
Ian Littlewood

About Ian Littlewood

Ian Littlewood is a long-time Altairian of nearly 20 years. He holds a degree in mechanical engineering and began work in engineering design on industrial and automotive machinery in 1991, then transitioned to CAE analysis — and HPC systems — while working primarily on car suspension components, cross car beams, engine mounts, and other projects for key global OEMs. This started a long history of using workload management tools and tuning application sets on HPC systems. At Altair, Ian has supported Altair HyperWorks™, where knowledge of desktop and HPC application tuning is critical, and managed Altair’s UK HPC footprint. He moved into workload management with Altair PBS Professional™ 15 years ago to support, implement, and architect customer solutions for major customers. After a period in sales, he transitioned to product management to bring his breadth of knowledge to bear in mapping customer requirements into the software development process of Altair PBS Works Suite.

Comments are closed.