For some time now, Facebook has been sharing the IT infrastructure technology it developed to run its own IT operations with the rest of the IT community via the Open Compute Project (OCP). At the OCP U.S. Summit 2017 conference this week, Facebook refreshed almost the entire OCP portfolio, including adding a much denser Tioga Pass compute server that employs dual-socket motherboards to provide more I/O bandwidth, a Bryce Canyon high-density storage server, and a Big Basin server optimized to run neural networks that are foundational to running deep learning algorithms.
At its core, the OCP platforms are part of an effort to flatten compute, storage and networking in a way that makes data centers from an IT infrastructure environment much denser. That, in turn, becomes key to running applications at web scale. But without tooling to manage those data centers, OCP as far as the average enterprise is concerned, is an interesting exercise in research and development.
This week Liqid, Inc. at the conference moved to make OCP a lot more relevant in the enterprise via the launch of the Liqid CI Platform, which provides the software needed to treat OCP platforms and other hardware as composable infrastructure that enterprise IT organizations can more easily configure and deploy as they see fit.
Sumit Puri, vice president of marketing for Liqid, says rather than having to standardize on specific sets of pre-integrated server platforms running a specific type of virtual machine software, the Liqid CI Platform unifies the management of bare-metal IT infrastructure via the Peripheral Component Interconnect Express (PCIe) bus in accordance with specifications outlined by Intel in its Intel Rack Design (RSD) architecture.
That approach, says Puri, means IT organizations can mix and match compute engines, storage and networks as they see fit, using open systems, without sacrificing any flexibility in terms of best-of-breed technologies. Perhaps best of all, the need for commercial virtualization software such as VMware to achieve that goal is eliminated altogether, says Puri. The Liqid CI Platform can be invoked via either a Liqid Command Center interface or RESTful application programming interface (API).
“We can configure a server in seconds,” says Puri.
In general, Puri says this is part of a larger trend toward the disaggregation of IT infrastructure resources inside the data center to achieve more flexibility. While many commercial IT infrastructure vendors are pushing for the adoption of converged and hyperconverged infrastructure to manage data centers at a higher level of abstraction, as far as the Liqid CI Platform is concerned those systems are just another type of infrastructure that can become part of a larger composable fabric.
There’s probably more debate today over the future direction of IT infrastructure in the data center than any time in recent memory. The issue that IT leaders might want to concern themselves with is finding the best way to leverage multiple IT infrastructure architectures without unnecessarily having to commit to one now, only to find out it’s been superseded a few short months later.