Azure N-Series VMs have hit general availability, much to the delight of technology enthusiasts across the globe who were waiting for the service to take a step past the preview stage. Azure N-Series VMs leverage the power of Nvidia’s GPUs to deliver accelerated computing to enterprises.
N-Series VMs use the all new Tesla line of cards from Nvidia. There are two instances in N-Series VMs, but to serve different purposes. They are: NC and NV.
Azure NC virtual machines: GPU compute
NC instances in Azure have been designed to deliver cloud-hosted GPU computing power to applications that require significant parallel processing of massive data sets. Deep learning, ray-traced rendering, geologic data — these are just a few of the data-intensive applications where cloud based GPU computing has tremendous scope.
Using Tesla K80 card along with its 4992 CUDA cores, NC instance is a dedicated parallel processing machine suitable for Big Data applications. But can it make you pancakes in the morning? No? Well, then, it still needs some upgrades!
Customers can use NC instances to manage training jobs that require deep learning, high-performance computing simulations, DNA sequencing simulations, CUDA accelerated jobs, and real-time data analytics. Powered by Nvidia’s K80 GPUs, Azure NC instances are ideally suited for all high-performance computing requirements, application acceleration, and AI workloads.
Note that NC instances are not meant for use with virtual desktops. So enterprises with Big Data hosted on-premises and on cloud can consider these instances, not the ones looking for computing power for virtualization.
NC instances are available in configurations such as 6 CPU cores and one Tesla K80 on the lower end, to 24 CPU cores with four Tesla K80s on the higher end. NC instances running in Windows are only beginning, and mostly concentrated in the eastern U.S., the south central (no, this has nothing to do with some gangster rap song by Ice Cube! You are thinking of something else!) U.S., Western Europe and Southeast Asia, priced between $0.66/hour and $3/hour.
Algorithmia, an open marketplace for algorithm development, has 30,000-plus developers using more than 2,500 custom algorithms. The organization is using on-demand GPU instances to deliver accelerated deep-learning algorithms and computation models to teams. It’s the Azure infrastructure flexibility that helps users meet the needs and demands of users.
Azure NV virtual machines: GPU visualization
For enterprises looking to adopt upcoming technologies that can enable the next level of virtualization, Azure’s NV instances are more attractive. These instances use Nvidia Tesla M60 cards, and are available in virtual machine configurations beginning with 6 CPU cores and 1 M60 GPU, to higher end configurations of 24 CPU cores with 6 M60 GPUs.
Note that currently NV instances are using pass through. So, it’s not possible to further slice the GPUs (that would be the equivalent of running VMs in your virtual machines). This doesn’t make the option any less attractive, though, because these virtual machines can be used as RDSH servers, or to host high-end workstation workloads.
NV instance pricing varies between $0.73/hour to $2.92/hour. The availability, like NC instances, is currently limited, but is expected to reach other regions soon.
Azure NV instances can deliver Nvidia GRID capabilities on demand. Engineers, scientists, and designers can use these instances for hardware-accelerating workstation applications. NV instance supports applications using OpenGL and DirectX.
Frame, an enterprise cloud platform, allows any kind of Windows software to be run in the cloud, and rendered on any browser. Frame is used as a platform in several institutions for delivering graphics intensive 3D applications in the form of Software as a Service.
Frame leverages Azure NV instances to provide an almost workstation like experience to all connected devices. Switching from NV6 configuration to NV24 configuration can be done in a split second, allowing users to quickly scale up and tackle complicated computing problems quickly. No, should I buy crunchy or creamy peanut butter is not a complicated problem! Well, unless your name is Napoleon Dynamite! Or Joey from friends!
The way forward
Azure is creating the right examples in the marketplace by tapping the power of GPUs for managing the workloads for modern enterprises. Azure has already become a critical part of the large taxonomy of cloud service providers that use GPUs:
- Infrastructure as a Service: Enterprise quality virtual machines (backed by GPUs), for virtualization of applications and workstations.
- Desktop as a Service: Enterprise grade GPU-accelerated virtual desktop infrastructure solution stacks, or customized offerings built on Azure NV series virtual machines.
- Independent software vendors who want to add the computing power of GPUs to their apps, leveraging Azure NV series to deliver their solutions backed by GPU on the cloud.
VDI ecosystem is transforming, with software vendors looking to deliver user experiences that replicate the powerful performance of physical hardware. Like any application you run on your laptop or personal computer, even the virtualized applications, right from the simplest clients to the most sophisticated ERP, need the computing power of GPUs to be significantly accelerated.
By integrating Azure N-Series and virtualization applications such as Citrix XenApp, ISVs can create solutions that truly leverage the power of GPU-backed cloud and deliver accelerated application user experiences in a virtualized environment. Azure N-series has the potential to empower all kinds of cloud-enabled virtualization services.
For graphics-intensive use cases, such as those in designing and movie production, the power of Azure N-Series virtual machines backed by Nvidia Tesla M60 GPUs is highly empowering (it probably cannot help out “Star Wars,” though; there has not been a decent “Star Wars” movie made in decades. “The Force Awakens” was pitiful!). This enables end users to get workstation class performance even while using high performance computing applications.
Strength of combined power
Azure N-Series is the biggest step forward towards the realization of the combined power of cloud, virtualization, and GPU accelerated computing. The number and scope of business use cases for Azure N-Series VMs for enterprises will continue to expand and will drive its proliferation.
Also, the emergence of vendors in the SaaS and DaaS space, either delivering Azure N-Series VM powered services, or delivering customized solutions that build upon Azure, will expand the business adoption. Enterprises with heavy computing requirements, large-scale virtualization requirements, and expanding scope of Big Data and analytics would do well to stay in sync with the advances in Azure N-Series VM services.
No, your kid’s Little League’s snack bar probably does not need this! Making scrumptious chili cheese nachos does not qualify as a heavy duty computing!
Photo credit: Microsoft