December 21, 2014

Scale-Out Storage Field Day?

It’s really difficult to scale storage!

Let me set the record straight: Tech Field Day events don’t have official themes. We don’t select companies based on what sort of technology they are selling or tell our presenters what to focus on. But it sure seems like we do sometimes!

At Storage Field Day 3 in April, it seemed like everyone was talking about flash caching. Delegates at that event saw the launch of Pernix Data and were briefed by SanDisk’s FlashSoft team, Marvell’s DragonFly group, NetApp’s flash caching concept, and even a surprise visit from Fusion-io, who announced their acquisition of NexGen storage there.

Having wrapped up Storage Field Day 4 this week, it seems that the theme was scaling storage. Delegates learned about scale-out storage from CloudByteCoho Data, Nimble Storage, Overland Storage,  Avere SystemsGridstoreOxygen Cloud, and Cleversafe. Of course, we also heard about storage caching from Proximal Data, Avere, and Virident. See, I told you there was no theme!

Essentially, we have five companies trying to scale out arrays while preserving existing client protocols and three companies using atypical client access methods:

  • CloudByte, Overland, Nimble, Coho, and Avere have developed clever ways of enabling existing storage protocols (iSCSI, NFS, and Fibre Channel) to access more than one storage system (node, array, target, filer, what have you) without knowing it. This is really difficult, since those protocols were never intended to work this way.
  • Gridstore, Cleversafe, and Oxygen Cloud all require different client protocols or drivers to enable scale-out access. This poses a challenge for IT staff who don’t want to abandon tried-and-true protocols, but all three are transparent to users and applications once everything is configured correctly.

I’ll briefly write up each of these scale-out storage solutions here on the blog. But this post started to get excessively lengthy so I decided to split it into three posts:

  1. Scaling Storage In Conventional Arrays
  2. Scaling Storage At The Client
  3. Scaling Storage In The Network

Cute kitten image by yours truly. That’s scrappy the nanocat!

  • Pingback: Scale-Out Storage Field Day?()

  • Pingback: Avere your Cloud | VDICLOUD.NL()

  • Peter Chang

    Great perspective Stephen. Scale-out storage and scale-out access are good differentiations.

    Another way to compare vendors is to increase scope and look at their core value propositions. From my perspective, storage vendors principally:

    1) make storage cheaper
    2) make storage more usable

    There are of course many many use cases a vendor can target for optimization. The best product depends on the desired combination of cost and usability for a specific project.

    For Oxygen, we believe users and application have moved to the cloud (outside the LAN). Our approach is to optimize for user and app access to storage from the cloud. “From” is the key optimization. Instead of replacing existing storage, simply enable the new access paradigm. Users get intelligent sync folders for easy access to any storage, application developers get a REST API to any storage, and IT administrators get to use ANY storage system.

    Peter Chang
    Founder & CEO
    Oxygen Cloud

  • Pingback: Server and Cloud Partner and Customer Solutions Team Blog()