I’ve been thinking about my last posting, that of the quagmire of licensing. I realized that I had only posted about one side of the dilemma. The licensing model of enterprise software can be a daunting process, and certainly a large financial drain on an organization. But that’s really only half the battle. What about equipment? The support/licensing model also, can be a huge problem.
Back when I worked in an enterprise, which was a very large one, we housed a mainframe in our California data center. This machine had been classified as End of Life, or outside its initial warranty period for years. The amount of money our organization paid for 3rd party support was a huge expense every year. Should we have chosen to use post-EOL support from the manufacturer, it would have been even greater expense. There’s no telling whether our support levels would have been better from the source support group, but we certainly would have spent quite a bit more. It’s been said that three years of support on the exiting platform could be enough in dollars to replace the device itself. One wonders if the same model exists in other categories. Large storage arrays, mini computers, X86 servers, and not to mention, the switch infrastructure.
Surely, augmenting the cost of replacing this equipment with new, and first-party supported hardware requires migration, which entails its own costs in manpower, and even training, should that be required. For example, the Cisco movement from Catalyst toward Nexus which happened a few years ago. Cisco placed a supportability deadline on Catalyst. There were two models presented. Either upgrade your catalyst environment to Nexus, or go to a third-party support model. This put the customer in a tough position. There were only three options. They needed to spend, in order to maintain compliance, and support on their infrastructure. The costs were substantial. They could have chosen to go without support, on viable equipment that was still quite functional, and run the risk of outages without support. Or, they could upgrade to Nexus. Needless to say, this draconian action caused a shakeup within that particular industry. Many organizations replaced their entire Cisco environment with other brands (Brocade, Aruba, Juniper, and a number of other manufacturers), others bit the proverbial bullet, and upgraded to Nexus. What that market became was far more complex, and very quickly. All these manufacturers have quality products, and I understand the rationale chosen by many of my customers, regardless of what direction they chose.
Storage presents a similar quandary. Replace all the storage with new, migrate to that new platform, ensure all connectivity either by existing fabric, or new, and train train train. Does that make sense in a market now comprised by vendors building on commodity equipment? Well, it can. Or alternately, the company can determine that the existing hardware provides all the stability, connectivity, IO, etc. that they require, and they can run it either supported, or pay the fees required with in-band or out of band support.
A model that I particularly like in the storage space is that originally marketed by Pure Storage, and now mimicked by many vendors, called by Pure: The Evergreen model. To start, when purchasing a Pure array, (M or X series), you purchase support alongside the equipment. Nothing beyond that is extra. Three years later, Pure will come back to your site, with two brand new x86 based head nodes (the key here is new. Processors and motherboards will have undergone major refreshes by then. And they’ll install these non-disruptively, essentially granting the organization a full array refresh. This will continue for as long as the device remains under support, effectively giving the organization a much longer lifecycle on this equipment. Should the need for disc grow, all that’s necessary is that the organization add to the existing support contract more disc to be covered. And, of course, as the unit is under support, should any drive failures occur, those get replaced as per the terms of the contract. In this way, the migration, replacement, and ultimately supportability of their storage array remains viable.
I foresee this type of model continuing, and extending outward as it really shows a customer centric model, solving the pain of replacement, learning curve, and continuity.