An enterprise committing to virtualization and cloud is a lot like a person committing to serious physical exercise. In both cases, it’s a question of eliminating excuses.

A personal trainer may ask: Do you want to become thinner? Stronger? More flexible? What about all three?

At first, “all three” may sound like too big of a commitment, even though it’s what most people really want. So the client might say “I just want to become thinner. The weights are just too intense, and I get intimidated by the flexible people in yoga classes.” And the trainer will just nod, and bide his time. He knows that the real answer is “I want all three, but only under the right circumstances.”

Are you ready for true virtualization?

A consultant guiding a service provider into virtualization may ask the same question: “Are you ready to jump in, do the work, and really transform yourself? Are you ready for true virtualization? Or just a piecemeal solution?”

Instead of getting thinner, stronger, and more flexible, the service provider is considering compute, memory, and storage, and whether a virtualized environment can address its needs in all three of those areas.

For hardware historians, two milestones will come to mind here: the launch of Intel® Xeon® Clovertown CPU in 2006 and the launch of Intel's Nehalem microarchitecture in 2009.

Consider the Intel® Xeon® Clovertown VT-x. The "VT" stands for "virtualization technology," meaning that the chip was made to work with a virtual environment. The chip also featured CPU offload, meaning that the CPU didn’t have to do all the computing on its own; some of it could be done by one or more graphics processing units.

For some—but not everyone—the Clovertown CPU was the sign that it was time to virtualize.

Three years later, Nehalem brought page tables, which map physical addresses to virtual addresses. This allowed for memory offload in virtualized environments. The combination of CPU and memory offload was enough to convince many—but, once again, not all—enterprises to get serious about virtualization.

Intel® Xeon® Scalable Processor removes obstacles to virtualization

Now, the Intel® Xeon® Scalable Processor has removed the last of the obstacles. It comes with a field-programmable gate array (FPGA), so that you can do the offload in a general-purpose chip. You don’t need a special FPGA somewhere else in the system to do that. And it’s got a non-volatile memory controller. So you don’t need a separate controller for implementing super-fast storage.

With memory offload, CPU offload, and fast storage all available, there is no longer a reason not to virtualize. There can't really be any technical objections to network functions virtualization.

Software-defined infrastructure, Intel® Rack Scale Design, and Intel® Xeon® Scalable Processor

By deploying applications on software-defined infrastructure, powered by the new Intel® Xeon® Scalable processor, all the pieces are in place to go all-in on cloud.

By pooling resources and using a smart hardware orchestrator, service providers can easily tailor virtual performance-optimized datacenters (PODs) for various cloud environments and applications. The enabler to achieve this flexible system is Intel® Rack Scale Design. It is an architecture that makes it possible to disaggregate the system and manage its lifecycle on a component level; thus, we cab always maximize the benefit of Moore’s law. Each component in the system has the potential to be on the technology forefront independently of the other components at any point in time. Put the components on a common optical interconnect and you’ll have a system that scales beyond the rack—not only inside the rack.

By combining Intel® Xeon® Scalable Processor with software-defined infrastructure and an optical interconnect, service providers have the performance, scale, and flexibility to successfully transform to a modern environment. Technology will not be the showstopper. We at Ericsson are right now deploying Ericsson Hyperscale Datacenter System 8000 based on Intel® Rack Scale Design on the new processor with some of the most demanding service providers on the market. The technology is available and there are no excuses not to take advantage of it.

Are you ready to hit the gym?

Read more about how Ericsson implements Intel® Xeon® Scalable Processor

 


Cloud Infrastructure

Henrik Bäckström

Henrik works as a Product Marketing Manager at Business Area Digital Services, focusing on cloud infrastructure products and network applications. He has worked for Ericsson since 1999, starting with product management and commercial management for fixed access before going into marketing for several core network and cloud offerings. Henrik has a MSc BA from Stockholm University.

Henrik Bäckström

Discussions