The phrase refers to infrastructure that an application can grab as necessary to meet its specific needs. The key is that compute, storage, and networking have no interdependencies, so that the application gets what it needs — and that this all happens automatically.
Synergy was the highlight of an otherwise hand-wavey keynote session, where HPE’s handful of product launches was skimmed in favor of high-level proclamations about the looming IT transformations that enterprises are facing. Key among them, executives said, is the need to adopt a hybrid model, blending on-premises IT with cloud-based applications.
HPE claims Synergy is the industry’s first true example of composable infrastructure. Then again, as The Next Platform points out, neither Synergy nor Cisco’s UCS M-series addresses one key dimension of flexibility: the ability to increase memory independently of CPUs.
Physically, Synergy looks like hyperconverged infrastructure. It’s a chassis to which modules of compute, storage, and networking can be added. (The networking piece can become a replacement for top-of-rack switches.)
As applications come into being, they would draw the necessary amount of computing, storage, and networking from the resource pool created by Synergy. When the application is terminated, that infrastructure would automatically be undone.
Being early, or possibly first, to this market, HPE is eager to set the boundaries for what counts as composable infrastructure. It has to support traditional bare-metal IT, not just virtualized environments, said Ric Lewis, HP ‘s senior vice president of converged infrastrucuture, during the keynote — an apparent swipe at VMware‘s software-defined data center (SDDC).
Composable infrastructure also has to be controllable through one API, Lewis said. “It can’t be a hierarchy of managers and managers.”
Synergy is due to ship in the second quarter of 2016.
Disclosure: Craig Matsumoto is attending HPE Discover as a paid speaker.