What's In Your Datacenter
There is a series of advertisements for Capital One aired in the US featuring Vikings talking about “more points” from their credit cards that asks “What’s in your wallet?” While they’re entertaining, I never understood what Vikings had to do with a credit card, other than perhaps both like to plunder unsuspecting innocents. Though in fairness, credit card issuers tend to just increase rates, while the Vikings enjoyed wholesale slaughter when they plundered, and took literally everything not nailed down.
But the question is valid in the modern day. Most people have enough credit cards, and read little of what is sent to them relating to those cards, that if you are like most people, you probably don’t know for certain what is in your wallet. What are the interest rates, current balances, and limits on all the cards you own? A surprising number of people cannot answer those questions – either because they have more cards than are readily memorized, or because they threw out all the paper that came with their billing statements that described changes to some or all of the above.
The author, with Lori and a collection of friends
This problem is very much true for modern IT departments also. I’ve talked in the past about getting the most out of the gear you have… Utilizing F5 BIG-IP LTM for load balancing, for example, is good since that’s where its roots are, but as a full-blown ADC, there is much more it can do that might save you the time and expense of buying another product. But there’s more. Some of the gear in your DC is redundant, some came in with a product, and you have never considered extending its use or eliminating it entirely. And that just might be costing your organization. It could be costing the organization a lot.
Let’s say you implemented a Virtual Desktop Infrastructure (VDI), for example, and the vendor that sold it to you sold a “preferred” ADC solution to go with it. While VDI is a heavy network use architecture, it should not use all the available resources on said ADC. So the question is “have you considered using it more broadly, and have you considered the use of other products that are deployed more broadly to solve the VDI ADC issues?” If every architectural solution you deploy comes with its own set of network components, you will soon have a network more complex than the minotaur's maze, and harder for operations staff to navigate. And seriously in the VDI case, if you’re virtualizing servers too, then getting specialized hardware just for VDI seems… Really very redundant (and why I picked VDI for an example, though other markets do the same thing).
We have steadily moved away from dedicated hardware for servers – particularly in cases where the servers were under-utilized – so purchasing different solutions for a single project or product seems like going backward. Network resources can be centralized and utilized across the organization – indeed, for a very long time they were. Only recently have some vendors started to delude themselves into believing that their product or solution is so special it needs its very own set of network resources. Consider for a moment the implications of “pay for the product, then pay us for this redundant infrastructure too” on your bottom line. Not pretty. And even if they give it to you free (I can think of a couple of vendors that have done this in the past either to increase apparent market share or to mask the cost of the infrastructure), you’re still paying for it one way or another. Nothing is free.
So if you have hardware sitting around that was sold as “special purpose”, and you’re using it just for one segment of your architecture, look around. Either you could use it more, or (more likely) you could use something else that does more than just the single task. Why pay repeatedly and introduce a bunch of different hardware/vendors into your growing network, when you can re-use or extend one set of gear to meet multiple needs. K.I.S.S. for those who recall that acronym.
So pay attention. Don’t settle for less because a software vendor is selling you hardware. Don’t purchase a solution you know could be used more globally unless you actually can use it more globally, and make sure that whatever hardware is in your datacenter is being utilized as much as you can/need to. And when evaluating new products, look at how you could fulfill needs from tools already in place or heterogeneous solutions before you go to buy special purpose. Save the budget. Waste not, want not and all.