Last May, @RayLucchesi and I met with Charles Fan on the Greybeards on Storage Podcast, which you can listen to here. It was a great conversation. I really enjoyed it and was looking forward to this conversation at the Tech Field Day event to see what progress has been made by these folks.
A bit of history: These guys, for the most part have come from the storage world, Charles particularly, was a critical part of VMware’s vSAN product, which is quite a pedigree. With the advent of Intel’s 3D Crosspoint product, in the Optane device, as it has been leveraged on the PMem space, rather than Intel’s storage product, MemVerge has taken the approach of capitalizing on what has become a very game-changing element of the compute side of the equation. As Optane Memory gives the architecture of server side computing a compelling boost in functionality.
Catch the Tech Field Day video here: https://t.co/zKuMr1PkQV
Over the past few years, I’ve discussed the expanding memory capacity on the server side, but MemVerge takes this approach to a level that hasn’t been achieved, with a full complement of software to allow for the functionality to be truly capitalized in the markets of Artificial Intelligence/Machine Learning, Database infrastructure, Virtualization, and High-Performance Computing. As more of the data, operating system and even application can load up into memory, there become fewer calls to the storage subsystem, which allows for levels of performance and efficiencies never seen before in the X86 world.
Not only can you load huge databases into memory, you’ve got amazing capacities to do things like take snapshots of databases instantaneously, and that backups for these applications, all driven by API’s become almost trivial tasks again achieving performance and functionality gains with power and speed, sitting in categories we’d only seen in movies prior to this.
The capacities for visual recognition and other highly demanding AI related tasks could change the game. As processors, and graphic processors become even more plentiful, the densities of these workhorse servers, and the sheer quantity of them geared toward these workloads needn’t be quite so enormous, yet far more performance oriented. I really look forward to seeing how the incorporation of these technologies into some of the projects I’ve been working become more efficient, more powerful, and more robust. I’m excited about the Memory Machine and where MemVerge will be taking it