We hope you’re taking it easy over the July 4 break this year, but if the hot topic at your socially distanced BBQ is data center architectures, you’ve come to the right place. IDC estimates the composable infrastructure market — which consists of appliances/platforms, software, and associated hardware drag — to be $751.5 million as of 2018, with expectations that it will approach $5 billion by 2023 with a CAGR of 58.2% by 2023.
Though other providers claim composability, they are unable to provide the kind of comprehensive, adaptive architecture required to meet the evolving demands of AI+ML, 5G/edge deployments, HPC, and other high-value applications. Liqid composable infrastructure solutions enable IT users to:
- Increase data agility, capacity & bandwidth in data center environments for which lightning fast, adaptive data manipulation is necessary;
- Leverage pools of hundreds of disaggregated GPU, FPGAs, CPUs, NVMe, & storage memory to adapt to surges in data activity, either on-demand or through automation;
- Multi-fabric capability enables IT users to share accelerator resources either up and down the rack via PCIe for maximum performance, but also across Ethernet and Infiniband for distance: another industry first;
- Deploy composable software across intelligent fabrics to disaggregate tightly converged resources and incorporate accelerators to scale at unprecedented volume.
The results are really something to celebrate. Liqid has collaborated with the tech industry’s leading companies to create one of the industry’s fastest GPU supercomputers with off-the-shelf components and to deliver the industry’s fastest single-socket servers with composable NVMe storage. We also worked with the University of Illinois at Chicago to deploy infrastructure that amplifies adaptive GPU computing performance for visually intensive scientific research and performance-oriented computing.