Even with my 20 years of experience working in technology, I am just a baby compared to some people I have encountered along the way.
Still, I sometimes think about the “old days” in IT. It was a time when data centers were dominated by mainframe systems and midrange monoliths like the AS/400. A time when virtualization was something done only on these big systems, and the only need for an x86 system was under your desk to access a mainframe when you didn’t have a classic thin client.
When my career in IT really began, I was working with systems like Windows NT and Exchange 5.5, and we were creating Web pages with this mysterious language called HTML that magically made things appear. Even in those days, there was this thought that there was no need for companies to be beholden to the largest vendors.
There was a notion/belief in the early 2000s that with the growth of VMware, eventually the data center would become a blend of multiple vendors and companies would be able to take back their own infrastructure, enabling them to control the way they supported their business. This transition marked the evolution from the traditional data center to the commodity-based data center, driven by business needs and created using software, rather than the monolith systems that big vendors charged exorbitant amounts for.
Fast-forward to 2016
Are we there yet? Not quite, but this could be the year. Most enterprises have nearly 90% of their systems virtualized on VMware, Xen or KVM and are managed by newer solutions like OpenStack. 2016 could finally be the year we see the commodity-based data center become reality.
To make this happen, there are so many components that need to be in the right place. We need to see x86 servers with enough power to support large database systems — check. We need the ability to choose between multiple vendors that all produce a similar-quality product, but differentiate themselves with their support and supply chain — check. We need to have networks that can handle multiple systems and the communication between them, without losing packets and without being based on a single large infrastructure — check. (Well, sorta . . . we will get to that one in a few.) Finally, we need software to drive all these systems and define the solutions, whether it is storage, compute or even networking.
We are just about there and are seeing an influx of companies that can provide these services, but can they unseat the incumbent large, publicly traded vendors that look to buy up all the small fish? That just may be the key to 2016 being the year of change for the data center we have all dreamed about since we walked away from worrying about what happens when the clock strikes midnight for the new millennium.
Over the next couple of blogs I will take a look at each of the categories and see if we are really as close as we think to 2016 being “The Year,” starting with the server and compute side.
Won the battle and the war
There is little doubt that the data center is now dominated by x86 server hardware. The hardware can run with multiple CPU manufacturers, and has more cores than we could imagine in the ’90s and memory levels that exceed what helped put a man on the moon. All of this hardware is going to support everything from the large single applications to scale out big data apps that let us analyze every picture that was posted on Instagram or Facebook about kittens in the last 24 hours.
Let the vendors compete for you
Most users probably have their preference now in server vendors, and it is all about the extras these companies provide. It is not about the components within the servers. Most network cards are from one or two companies, and most processors are from one of two companies — you may have hard drives from multiple firms, but overall it’s just different-shaped, bent metal with a different label on the front.
The true differentiator for vendor choice comes in the form of supply chain and parts availability. The supply chain has begun to collapse around fewer global vendors, making it easier to price match between them. If you are only looking for regional or local support, that too has limited choice, and all the vendors know it. The choice is now truly in the hands of the consumer and, if planned out, can result in a race to the bottom on price, along with increased “hand-holding” when the equipment arrives.