Blog

NVM Express Computational Storage at Storage Developers’ Conference 2018

Every year SNIA holds an event called Storage Developers’ Conference in San Jose. This year’s event was, as usual, an excellent technical conference attended by a Who’s Who of the storage architect world. As well as having excellent technical tracks this event is renowned for its networking with plenty of opportunities for hallway and bar conversations.

Eideticom At SDC 2018

Eidietcom had a great SDC2018 with three technical talks and two Bird’s of a Feather (BoF) sessions. Our talks and sessions touched on two themes that were all the rage at SDC2018:


  • NVM Express for Computational Storage: Eideticom has been using NVM Express for Computational Storage for some time now via our NoLoad product. We presented some use-cases of NoLoad around NVMe over Fabrics and computation. Also, at SDC2018 there was a lot of discussion around how we add some of this to the NVM Express Standard. This is something SNIA is driving via its Computational Storage Technical Working Group.
  • Peer-2-Peer Data Movement via the P2PDMA Framework: In order to aid NVM Express Computational Storage it helps to make it easier to move data in and out of NVMe controllers. At Eideticom we have been pushing a framework called P2PDMA into the Linux kernel to assist with this. There was a lot of interest in the status of P2PDMA and how PCIe driver writers can update their code to leverage this new framework.

conference

Figure 1: Full house for the Computational Storage BoF Session at SDC2018. People are interested in seeing how best to standardize Computational Storage so we can build vendor-neutral systems, update operating systems and create an ecosystem.


fellas

Figure 2: My interview with Intel Data Center Builders discussing what I saw at SDC2018 and what Eideticom presented. The video for this is here.

Where Next?

  • The videos and slides from SDC2018 should be online here.
  • The SNIA Computational Storage TWG continues to move forward. You can track progress here.
  • The P2PDMA framework continues to move upstream. The code for this work can always be found here.