We appreciate HP and their push to make composability a more mainstream solution. But is Synergy truly composable? We think it falls short in some critical areas to enable true composability.
First: disaggregation is a key hallmark to true composability. You should be able to scale SSDs, CPU's, GPUs, FPGAs all independently. Scaling based upon converged boxes isn't optimal.
Second: bare metal composability. Composability without bare metal is more along the lines of hyper-convergence. Doing things like virtualization or adding a hyper-visor layer is not true composability. The only way to achieve true bare metal composability with the physical data center resources that are available today is with PCIe.
That's another key difference between LIQID and HPE. They choose an Ethernet based fabric and we choose a PCIe based fabric. By using a PCIe based fabric, we're able to do bare metal composability where they are not. Plus, there are no Ethernet SSDs. There are no Ethernet GPUs. So if you're going to compose with Ethernet, you're not going to be able to do it at the bare metal. The last delta is around closed systems versus open systems.
LIQID is not only software agnostic, it's hardware agnostic. We work with a variety of hardware vendors. So if a customer has an environment where they have a variety of hardware in play, LIQID can make that environment a (truly) dynamic composable environment with different hardware types. Once you've gone down the Synergy path, the only hardware available to you is from that vendor.