Cloud-based engineering is transforming manufacturing
High performance computing (HPC) has given manufacturers the ability to perform with digital simulation what was once only possible with physical prototypes. These transformative technologies are empowering designers and engineers to visualize real-time product and factory performance from anywhere, simulate and replicate physical products, incorporate feedback from IoT-enabled products in the field and ultimately iterate faster on product design to accelerate innovation and improve speed-to-market.
High performance computing resources were once only accessible to large customers operating with their own on-premise resources and datacenters, and were often cost prohibitive to smaller customers. Now, with HPC resources available in the cloud, any business can experience the benefits these resources provide, packaged with the additional business value inherent to the cloud model.
As the market for HPC resources grows, however, how do you ensure you get the best possible return on your investment? We have identified 5 crucial questions you need to answer when developing a cloud strategy incorporating HPC technologies.
Is it comprehensive?
First and foremost, engineers will demand that they are supplied with capacity for all their compute-intensive workloads, regardless of simulation type or dataset size. They will also need to visualize those results without having to download, hence you also need GPU virtual machines (VMs). You also need to have high speed network connectivity to ensure high-throughput and low latency. In addition to a hyperscale global network, you also need a globally dispersed hyperscale cloud.
Is it scalable?
Scalability is another crucial component to any effective cloud solution and is one of the areas where the value and advantages of cloud computing are seen most readily. To take advantage of the shift to the cloud, you need to know that your cloud HPC resources can scale up and down easily to accommodate shifting business requirements. In doing so, you ensure that you only pay for what you use and avoid excess spending on compute capacity that you don’t need. Even if you have already invested in on-premise HPC infrastructure, you can still leverage the cloud to accommodate spikes in resource demand—a far more immediate and agile alternative to investing in more on-premises infrastructure that might take months to deliver. You should also ensure that your HPC cloud instance enables you to easily shift existing applications, scripts and tools to the cloud to benefit from more compute power on-demand.
Is it flexible?
The modern worker is mobile and device-agnostic. When investing in cloud HPC resources, you need the flexibility to connect from any device or operating system, be it iOS, Android, or Windows. Does your provider allow you to run workloads on both Windows and Linux? Do they enable you to experiment and run training jobs on any solver or framework? Are your design optimization engines supported? By ensuring that your cloud HPC solution is flexible enough to integrate with your existing tools and devices, you ensure that your designers and engineers can continue to work with the tools they are used to, but unshackled by capacity constraints on the cloud.
Is it elastic?
Your cloud solution should be elastic enough to adapt to peaks and valleys in demand by provisioning and de-provisioning resources accordingly. You want to avoid over-provisioning resources and pay only for what you use. Moreover, your cloud solution should ensure that VM provisioning is automated and scales up efficiently and quickly to maximize the proportionate spend on compute. At the end of the day, you want to ensure you are paying for compute, not provisioning.
Is it secure?
Designers and engineers are being called upon to collaborate more than ever, working across geographies while also collaborating across the supply chain with 3rd party vendors and OEMs. Given this paradigm, it is necessary to ensure that valuable IP is maintained in a central, secure location—something that is hard to do when sending screenshots over email. You need to ensure that your cloud provider will maintain your valuable IP securely in a central location enabled with controls to ensure that access is limited appropriately. Further, it pays to invest in a provider with a large geographical footprint to insure your data and compute resources are compliant with local restrictions and regulations.
Get Started Now – HPC with Microsoft Azure
Microsoft offers the most powerful, scalable, flexible, elastic, and secure cloud platform for HPC in the marketplace. Azure bundles NVIDIA GRID licensing with compute to enable customers to run GPU-powered applications on virtual machines. Azure also provides RDMA/InfiniBand high speed networking capabilities for high throughput and low latency that enables multi-node scaling of tightly coupled simulations. Plus, Azure Batch enables you to scale up and down efficiently directly from your applications to accommodate shifting business requirements, ensuring you pay for what you use, maximizing the amount of compute for your money, and minimizing the time and costs spent on provisioning.
With the ability to connect from any laptop or device, Azure empowers your designers and engineers to innovate on the go. Microsoft can also support workloads on Linux and Windows—offering native Linux clusters and several different Linux distributions. If you already have a Linux cluster, you can even redeploy it on Microsoft’s infrastructure. Additionally, Microsoft Azure enables you to maintain the right level of compute resources to meet demand, providing mature and proven orchestration capabilities for provisioning and deprovisioning VMs between on-premise and cloud clusters—a key differentiator from other major cloud providers.
Microsoft has also strived to make security and customer IP protection a priority at every step—from code development to incident response. The Security Development Lifecycle (SDL) addresses security at every development phase from initial planning to launch, while Operational Security Assurance (OSA) builds on SDL knowledge and processes to provide a framework to ensure secure operations throughout the lifecycle of cloud-based services. Microsoft Azure Security Center makes Microsoft Azure the only public cloud platform to offer continuous security-health monitoring.
The Azure Active Directory (AD) supports single sign-on at the individual VM level, thereby enabling customers to use the exact same security profiles that they use for on-premise engineering. Azure AD also includes a full suite of identity management capabilities including multi-factor authentication, device registration, self-service password management, self-service group management, privileged account management, role based access control, application usage monitoring, rich auditing and security monitoring and alerting. These capabilities can help secure cloud based applications, streamline IT processes, cut costs and help ensure that corporate compliance goals are met.
With 42 regions and counting spread globally, Microsoft is best equipped to allow you to choose where to run your applications and ensure your operations are compliant with local regulations.
Are you ready to unlock the full value of Big Compute for your business? Learn more about our Big Compute solution portfolio today.
