Making the Composable Data Center a Reality – Liqid Briefing Note

For the Software Defined Data Center (SDDC) to deliver its full potential it is necessary to bring together resources like CPUs, GPUs, storage and network into a dynamic pool from which applications or services pull those resources as needed. Then once that application or service finishes with the resources, it should be able to return those resources to the pool so they are available for use by another application or resource.

The concept of a dynamic pool of resources is known as composable infrastructure. In it, applications can pull resources dynamically on the fly as they need them and then return them once they finish. A data center that leverages composable infrastructure will see a reduction in physical hardware purchases, data center floor space, power and cooling while being able to always meet peak demand. Composable infrastructure is the cloud taken to its logical conclusion, and cloud providers as well as traditional enterprises data centers can leverage the technology.

While the names have been different, there have been attempts at delivering similar capabilities in the past. There were two problems with these prior attempts. First, there wasn’t a universally adopted high-speed network that supported direct connectivity to all the components. Now there is. PCIe has connectivity to storage (via NVMe Flash), Compute, GPU/Coprocessors, and networking. PCIe provides a global and universally accepted standard for accessing these components.

The second problem was the software either didn’t exist or was very basic in its functionality. But this has changed. Companies like Liqid are delivering a PCIe Fabric Switch that delivers full disaggregation and the software required to allocate and de-allocate resources on the fly.

Introducing Liqid

Liqid provides a PCIe switch that creates a PCIe Fabric, interconnecting networking, storage, GPUs, and CPUs. The switch is expandable to 24 PCIe ports, 192GB/s of bandwidth and 150ns of latency. Switches can be daisy chained together for expansion and N+1 failover. Storage and GPU expansion is accomplished via PCIe shelves that contain either NVMe flash drives or GPUs.

The PCIe switch comes with software, The Liqid Command Center, which enables the dynamic provisioning of bare-metal machines from disaggregated pools of physical resources. When a bare metal system is created the administrator uses Command Center to select the number of CPU cores it wants to use as well as the number of drives, network adapters and GPUs that they want to allocate to the new bare-metal system. From there the administrator can install the operating system and begin using the system.

Who Needs Liqid?

The most obvious candidates for using Liqid are cloud providers and hyper-scalers. The potential for better resource allocation and resource allocation mobility should give them the ability to meet their customers’ performance expectations better and of course, it should allow them to drive down cost.

A primary use case for Liqid is helping the organization preparing for augmenting Artificial (AI), Internet of Things (IoT), and Edge environments. Each of these environments often take full advantage of parallel compute capabilities of GPUs. They also are data intensive with unprecedented flows of unstructured data. Edge environments are attracted as well to help meet their demand for limited footprints.

The use case is not limited to cloud providers though. Enterprises stand to see similar gains for much the same reason. The ability to allocate and deallocate resources at a very granular level should deliver the same cost saving benefits to the traditional enterprise.

StorageSwiss Take

A composable data center is the software defined data center endgame. The ability to have all the data center resources sitting on a digital shelf to be pulled and returned as needed is the ultimate in data center efficiency. To make that endgame a reality though requires the right hardware configuration as well as orchestrating software. The release of NVMe flash drives and the demand for PCIe based GPUs and co-processors have made the hardware component practical. Liqid, through a combination of their switch and software orchestration, promise to complete the picture, enabling data center to reach new levels of efficiency while better keeping pace with user demand.

George Crump is the Chief Marketing Officer at VergeIO, the leader in Ultraconverged Infrastructure. Prior to VergeIO he was Chief Product Strategist at StorONE. Before assuming roles with innovative technology vendors, George spent almost 14 years as the founder and lead analyst at Storage Switzerland. In his spare time, he continues to write blogs on Storage Switzerland to educate IT professionals on all aspects of data center storage. He is the primary contributor to Storage Switzerland and is a heavily sought-after public speaker. With over 30 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS, SAN, Virtualization, Cloud, and Enterprise Flash. Before founding Storage Switzerland, he was CTO at one of the nation's largest storage integrators, where he was in charge of technology testing, integration, and product selection.

Tagged with: , , , , , , ,
Posted in Briefing Note

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 25.5K other subscribers
Blog Stats
  • 1,939,356 views