“Convergence” is a hot topic today but, like so many buzzwords, the term takes on many different meanings. But let’s consider converged infrastructure for a moment – systems that seamlessly combine server and storage. A variety of offerings can be called “converged”, from “stacks” of equipment curated by major vendors and resellers to integrated boxes with everything inside. Scale Computing is the latest company to join the ranks of the latter, introducing a “hyper-converged” solution combining a server hypervisor and scale-out storage software in a simple 1U chassis.
Defining Hyper-Convergence
In order properly to differentiate what companies like Scale Computing are offering, it is necessary to carve off a bit of the “converged” market. For this, I will steal a term from Stevie Chambers, who should know a thing or two about convergence since he works in the CTO office at VCE, the joint venture between Cisco, EMC, VMware, and Intel that sells the vBlock stack.
Chambers discusses a spectrum of unified or converged solutions, from specifications to stacks to boxes. And he places three companies (Nutanix, Scale Computing, and SimpliVity) at the end of the spectrum he calls “Hyper-Converged”. All three companies have introduced offerings that are similar in concept:
- A one-box design with (virtual) server and storage combined into a single chassis
- A scale-out solution that “goes big” by adding more simple nodes rather than (necessarily) bigger or faster ones
- Server-type hardware rather than a blades-plus-disk-shelf concept
- A unified management interface for server, storage, and hypervisor
Each of the three hyper-convergence companies approaches the problem in different ways, however:
- Nutanix runs entirely as virtual machines on a 4-blade “hyperscale server”, with a Virtual Storage Appliance on the VMware hypervisor running scale-out storage software based on the Google Filesystem (the company’s technical roots)
- Scale Computing runs virtual machines on 1U hardware in the Linux KVM hypervisor running “above” their scale-out storage solution, which is based on IBM’s GPFS
- SimpliVity created a scale-out storage array, complete with hardware offload for intensive tasks like deduplication and compression, that can also host virtual servers or support standalone non-converged servers
Unlike Nutanix and SimpliVity, which aim at the heart of the mainstream IT datacenter, Scale Computing is focused on small and mid-sized businesses. This translates into dramatically different price points, with Scale’s HC3 starting under $30k.
Seriously, go read Stevie’s blog post!
What’s in a Name?
When I first heard of Scale Computing, I was surprised by their name. The company had introduced a scale-out storage array competitive with mainstream offerings from the big names in the industry (EMC, NetApp, HDS, etc). Although the “Scale” side made sense, given their use of IBM’s GPFS technology, I questioned the “Computing” angle. After all, storage and computing are two different things!
But it soon became clear that this little Indiana startup had different goals in mind. They began to focus on small and mid-sized businesses who wanted big features at lower cost. And, with a wink and a nod, my contacts at Scale indicated they would be delivering a compute offering as well. Over the last year, I’ve kept up to date on their emerging compute product and waited for the day it was officially unveiled.
The nature of Scale’s HC3 offering is revealed by a look at their architecture. The company chose to use the KVM hypervisor for a few reasons:
- They could customize this GPL hypervisor and its drivers as needed
- KVM isn’t tied to a corporate giant who might see Scale as a competitor rather than ecosystem partner
- KVM is free (as in beer) so Scale can offer a lower-cost solution without a “VMware tax”
Scale is betting that their non-VMware offering can deliver an experience competitive to what a user of that company’s Standard-licensed vSphere would get. This includes flexible provisioning, mobility of virtual machines, and integrated management of the system. Scale also includes all the features of their scale-out storage array, but these are more thoroughly integrated than is possible for VMware-based solutions, which relegate storage management to a separate tab in vCenter.
But Scale is under no illusions that they are going to “knock off” the 500 lb gorilla that is VMware. No one has yet come up with a competitive solution for large enterprise customers, even though some big names (Microsoft, Citrix, and Oracle to name three) are hard at work doing just that. Smaller shops, who are more sensitive to cost and lack skilled IT specialists, are more likely to look for alternatives to VMware than large corporations.
Stephen’s Stance
I’ve met many small companies that just can’t justify the financial investment required to deploy a killer VMware vSphere solution. Some try to patch together a freebie based on ESXi, while others look to Microsoft and Citrix. These shops will welcome a new alternative like Scale Computing’s HC3 solution without being too put off by the absence of VMware. It’s a big market, and Scale has already attracted hundreds of customers for its storage solution. I’m betting many more will come on board now that HC3 is available!
Disclaimer: I do not sell blog posts, and this piece was not paid for or promised to Scale Computing. I found their solution interesting enough to warrant an honest writeup, and would have written about it regardless. But I must disclose that Scale Computing hired my company, Foskett Services, to help with the launch of HC3. This includes organizing pre-launch briefings, prime sponsorship and participation in Tech Field Day’s VMworld Roundtable video series, an upcoming webinar, and a participation in November’s Next-Generation Storage Symposium.