April 2, 2014
My marketing guy was discussing his golf swing the other day. He said that it had “more moving parts than the average car engine”. In case you are wondering, this is not a good thing since it leads to an errancy in shot making that creates an unsafe environment for one’s fellow golf aficionados. When you think about it, don’t we all subscribe to the theory that more moving parts equals more things to go wrong? And isn’t the thing that goes wrong always the hardest one to reach? Data centers are not immune to this whole “more parts bad” paradigm, but no one seems to care until something goes wrong—then everybody cares, a lot. I guess the obvious question is, “since everybody knows and agrees that this is the case, then why don’t they design facilities with fewer parts and that are easier to fix and maintain?”
Well yes Virginia, that is a good question, and the answer is that most data centers are not designed with operator efficiency in mind. This isn’t to say that most data center providers don’t try to design their offerings to be highly reliable. After all, if they didn’t why would there be so many Tier III+ facilities out there? This, of course, is a good thing since everyone wants their data center to operate the vast majority of the time. Unfortunately, this same level of concern regarding data center operation often does not extend to simplifying the ability of facility engineers and techs to perform routine maintenance or diagnose and fix problems when they arise. It’s kind of like they’ve adopted an attitude that says, “Hey we designed it to operate 99.995% of the time, after that, you’re on your own pal”.
When you think about it, how difficult is it for a provider to design a data center that cuts down on the moving parts and make it easier to maintain? Heck, even the car companies caught on to that by doing simple things like color-coding the dip sticks. Why not color code and label the conduit throughout the facility to make diagnosing and fixing a problem easier? Not only would this be an operational enhancement, you have to admit it would be pretty darn stylish as well.
The mechanical systems of the facility also seem like a good place to focus on to enhance operational friendliness. For example, don’t build it in the main floor. I know of one modular provider that has multiple sets of fans below the module itself. I’m sure their facility guys curse this design “anomaly” every time they have to go spelunking to perform some type of maintenance.
Front access to all major components is probably something that most folks could get behind. Forcing a tech to perform a variety of contortions that would have Gumby throwing in the towel isn’t just bad design, it’s just plain mean. When you’ve got a service affecting problem, the words you want to hear from the engineer attempting fix it are not “I can’t feel my legs anymore”.
I find it interesting that so little interest is paid to the actual day-to-day operation of the data center in the design process of most providers. Although no one wants a service interrupting event to occur, we all know that that they will and do. I have to wonder if this lack of operational efficiency in the design of many solutions reflects a lack of customer focus or simple arrogance on the part of providers. Of course, end users must share in the blame for these inefficiencies in design, since if they don’t make the ease of operations and performing maintenance a priority in their vendor decisions, they have to live with the ramifications of their decision. Perhaps this is due to a lack of awareness between both parties, since many times these flaws are only discovered with the occurrence of an actual event—maintenance or outage. I guess both providers and end users will have to begin thinking more proactively about the ease of operations. My marketing guy, for example, now yells “Fore” before he tees off. His fellow golfers appreciate this foresight, and I think a lot of end users should demand the same when making their next data center decision.