[UPDATE]: I had a further chat with SolidFire and I made a correction to the original post. Looks like, I’ve totally misunderstood a statement during the original session.
Last week I met up with SolidFire once again, and I’ve been wondering about their potential role in the enterprise space. If you’re a frequent reader of my blog you’re probably familiar with my POV about them… but there is something more that should be taken into consideration even though you might not be a huge enterprise or a CSP.
The company is small
and they don’t want to be competing directly with names like EMC or NetApp (just to mention the two most important vendors). and they prefer playing it safe and are concentrating their efforts where they have a clear advantage: huge cloud deployments (CSPs), for example. But, I think that their potential in the medium enterprise is huge… although not in every aspect but, in some circumstances their architecture, and vision, can make the difference in highly virtualized and private enterprise cloud deployments.
Not that big, after all
First of all they can start out relatively small. I know their message is all about scale-out, millions of IOPS, thousands of VMs, and so on… the reality however is somewhat different: their smallest configuration now starts with 4 nodes (which means approximately 200K IOPS and 48TB of usable space). Even though the array design is not meant to work in such small configurations (loosing a node means 25% of the entire performance and space), the tradeoff is acceptable in many real-world cases and in most mid-sized enterprise environments. BTW, if you compare a 4 node cluster with a dual controller configuration, at least from the performance POV, in the event of a controller failure only 25% of controllers is lost instead of 50 (and this is the worst scenario).
Good for private clouds
Solidfire, as they themselves admit, is not primarily built to serve one single specific workload (like a single huge DB, for example), but rather it’s designed to serve many different VMs (and workloads) at the same time and, thanks to the QoS features, without performance impact on each other.
The product also supports many different cloud orchestration platforms, which translates into freedom of choice and the development team is a strong contributor to the Openstack project.
One thing that I particularly liked is the “Openstack agile infrastructure”: A pre-validated reference design developed to deploy an OpenStack infrastructure using SolidFire’s storage system, Dell servers and networking, and Red Hat Enterprise Linux OpenStack distribution.
If Solidfire’s words can be trusted, this infrastructure could be up and running in an hour, and if so, it would be a great achievement!
This is the kind of result enterprises like the most, and thinking about deploying an Openstack infrastructure almost out-of-the-box could be very interesting for an enterprise looking into the private cloud.
I can understand why Solidfire is not looking for direct competition in the enterprise space, at least not from the marketing point of view. I also understand that their product, to be really efficient and effective, has to be deployed in larger configurations. And last but not least, I understand that Solidfire array is designed to manage several workloads at the same time and not just a few.
All good points indeed, but nonetheless they don’t arouse my curiosity or interest: I think there is potential which hasn’t been exploited yet by Solidfire. Solidifre could be a really interesting solution for enterprises looking for something new in the high performance virtualization/private-cloud storage space. Even if those enterprises aren’t that large…
Disclaimer: I was invited to this meeting by Condor Consulting Group and they paid for travel and accommodation, I have not been compensated for my time and am not obliged to blog. Furthermore, the content is not reviewed, approved or published by any other person than the Juku team.