Is hyperconverged becoming the new lock-in?

A few years ago backup appliances became all the rage, simplified super-speedy systems which the customer could stick into their datacentre and have up and running in a few minutes. Like these appliances, Hyperconverged currently seems to be all the rage – you get your Virtualised infrastructure from a single vendor who integrate everything you need, hypervisor, disk system, SAN, network, and all in a lovely scale-out environment.

What’s not to like?

As with traditional storage, the increase in software’s significance over hardware is there to be seen. The hyperconverged appliance, like the backup appliance, is just another generic x86 server stack with a vendor’s logo on the facia. Which must beg the question:

When so many other areas in the datacentre are going Software Defined and are based on generic x86 servers, why am I expected to buy in to a specific vendor’s application stack?

There are clear risks, primarily of vendor lock-in as can be seen with many backup appliances, you may be using generic hardware, but can that hardware inter-operate with hardware from other vendors, under the same software stack? All too often the answer is at best ambiguous, often a flat out “no”. There may also be limitations, are you stuck on-prem? can you move to and from the cloud? Do your existing software vendors support the hyperconverged stack?

So, I notice that IBM don’t have a hyperconverged appliance, why are you banging on about this?

IBM do have a hyperconverged solution, it’s Spectrum Accelerate running under VMware. You can build it yourself or purchase a pre-built version from Supermicro. A fully software defined solution, which eliminates hardware vendor lock-in. Spectrum Accelerate is the software portion of the XIV array, which brings up some interesting possible use cases. Not only do you have the ability to run a software defined hyperconverged stack, based on your own hardware choice (a compelling use case in and of itself) but you also have the ability to:

·        Replicate your primary XIV array to/from the cloud, allowing an on-site infrastructure to be moved seamlessly in and out of cloud hosted service, such as IBM Bluemix/Softlayer.

·        Operate a cloud based DR site based on exactly the same systems as your Production site.

·        Use your pre-existing backup software such as Spectrum Protect to control cloud hosted snapshots in exactly the same way as on-prem snapshots.

·        Use a virtualisation product such as Spectrum Virtualise to extend your on-prem storage into the cloud.

·        Use monitoring software such as Spectrum Control to manage on-prem and cloud arrays.

All in all, why would you want a niche manufacturer’s software defined storage system, in a potentially closed hardware stack and a disk subsystem of dubious pedigree, when you can have a Software Defined version with an Enterprise-class disk subsystem like Spectrum Accelerate, which has a long track record as XIV and excellent support from various software providers?