There is a conversation about the implications of data center convergence between the people that staff data centers today and the senior executives that foot the bill for those data centers that a lot of people are doing their best to scrupulously avoid.
Recently, the topic of data center convergence was the subject of some ridicule in a recent Dilbert cartoon, which probably served to raise more interest in a topic that most business executives have not heard that much about yet.
Data center convergence is a buzzword being used to describe a new class of integrated servers that leverage virtualization to combine processor, networking and storage technologies into a common platform. The basic idea is that rather than having separate servers, switches and storage arrays that need to be independently managed, all of these devices should be elements of a single commonly managed system.
You might well ask what has taken so long to do something that at first blush that might seem intuitively obvious, but it's only recently that we've had the 10 Gigabit Ethernet bandwidth, virtualization software and more energy-efficient processors and blade servers required to make these systems possible. The leading examples of these types of systems are the Unified Computing System (UCS) from Cisco, the BladeSystem Matrix from Hewlett-Packard and the Intelicloud 360 from Intelicloud Technology. Similarly, Dell has been promoting the concept of data center convergence via a partnership with Egenera, IBM has been touting a unified approach to the data center under a New Enterprise Data Center initiative and Sun is in the midst of a Project Crossbow effort that is anchored around virtualization software being developed for Solaris.
Right now, each of these vendors is making a difficult argument during these challenging economic times. Basically, they are telling customers that of if they spent a few million dollars today on creating a new data center architecture, they can have 10 times that amount in a couple of years. At a time when capital investment dollars are tight, this is a tough project to get off the ground. But as the same time, the fact that IT takes up such a huge percentage of the capital budget -- along with the high cost of IT labor -- is driving senior managers to starting thinking about the implications of data center convergence.
The two most attractive things about data center convergence are that firstly it reduces the actual amount of equipment in the data center. This in not only lowers the total cost of the gear, it reduces the amount of power required. Secondarily, these systems allow all the servers, networks and storage assets to be managed via a single management interface. That not only reduces the number of systems management tools that have to be acquired and mastered, it also potentially reduces the actual number of IT people required to manage systems in the data center.
It's generally accepted that about three quarters of the typical IT budget goes to simply keeping the systems running. The vast majority of that money is spent on the expensive IT specialists required to run separate server, network and storage architectures. The promise of unified computing in the data center is that it can substantially reduce the total costs of enterprise computing by simplifying systems management and reducing the number of people required to run a data center.
Naturally, vendors are not overly anxious to alienate the very IT staff that so many depend on to advocate the acquisition of their products. But at the same time, they are all keenly aware that because such a huge percentage of the IT budget goes to maintaining existing systems, the amount of money available to buy new systems is constrained. So the basic economic theory they are now working under is that by reducing the total cost of maintaining systems, there will be more money available to buy more systems.
Potentially, standing in the way of this fundamental shift in enterprise technology is the vested interest of the server, network and storage specialists on the IT staff. The fact of the matter is that some form of convergence in the data center is inevitable. The pace at which it happens depends very much on the ability of the senior leadership of the IT organization to repurpose IT specialists who have been primarily focused on maintenance tasks toward activities that actually add more value to the business. That's an extremely difficult conversation to have with people who would naturally see data center convergence as a threat to their livelihoods. And that's doubly true when you think about how bad most organizations are about retraining workers to give them new skills as the business evolves.
But there are literally thousands of ways that IT can be used more effectively. So the good news is that instead of spending all our time holding IT together, we might actually be reaching a point where we actually have the time to do something about if only we're willing to start the conversation.
So the question is, how hard is your organization thinking about data center convergence?