We’ve seen some fairly interesting announcements from some high profile companies such as Digital Realty Trust and Microsoft that they are using their own flavor of DCIM to manage their data centers. I caught up with Soeren Brogaard Jensen, VP Enterprise Software Solutions at Schneider Electric and asked him what sort of advice he would offer to other companies that were considering a data center management software implementation, especially if they were thinking of building their own DCIM in-house?
Soeren said that with any new software technology adoption there is always the choice of building it yourself: “You saw it earlier on with companies running their organization on MS Excel – it can be done very early on along the adoption curve. In DCIM the exact same thing is happening”.
The technology is also an aid in itself – especially the fact that a lot software applications are built in a modular fashion and use plain English for programming. Soeren said, “There are a lot of really interesting frameworks that are available now, that when you first get into them you can get very fast, very quick, and you can get excited.”
However they need to consider the long term implications. “What I would encourage people to look at is a couple of things: I would probably get them try to look ahead, maybe a year or two, past the first couple of releases, where things are great and exciting and you show a lot of progress. When you get into maintenance and you get into more data volume, you get into scalability. What looks easy when you first get going with a few hundred racks can potentially become complex when you get into sort of more 1000’s of racks and data”.
There’s also the business of core competence – or sticking to the knitting. Soeren said, “Also think about is software my core business? It could very well be that that’s what you want to develop, but the frameworks and the platforms that are available to do use, tend to be fairly generic by nature because they can be used for many things. Most of the DCIM tools that we see from vendors today have been purpose built, there is actually engineering that went into making them suited for the DCIM category”.
So the fact that generic tools can be adapted for simple tasks may not make them applicable as the requirement grows. In these respects the ongoing cost of software support and adaptation may make them uneconomic in the long run. There;s also the question of what business you’re in. Soeren concluded, “You can easily underestimate the value and the R&D time that went into them versus the generic tools. Purpose built tools can be fantastic for the immediate sort of proving that this is an area that I want to go into, once you’ve proven that this makes sense for you, take a step back, look at the frameworks and the purpose built tools out there and make your decision for the long run, ideally are you in the business of running a data center or are you in the business of selling software?”
10 years ago
Rolling your own enterprise-class software that has direct impact to the business strategically is usually a losing battle. It boils down to understanding your own core competencies. Sure I could cut down a tree and make a chair, but that is not MY core competency. I recognize that. The chair would likely be under styled, poorly constructed, take a long time to build, and ultimately not meet my needs for a set of chairs which all behave the same over the long term. There are furniture makers that specialize in chairs, and can do it better than me at any scale, and probably at a lower cost when everything is factored in.
My point is that building your own DCIM sounds easy at first, but usually just becomes a distraction to the core business. With so many changes happening in the data center, why would anyone choose to devote resources and time to building something that will be a support nightmare, have scale and usability issues constantly, and simply be a poor business choice over the long time? Would Finance build their own Financials package? Never!. So why do many data center professionals feel that building their core business management software is easy? I think this is a artifact of the past 30 years of some pretty amazing and resourceful behaviours. Look at all the SCRIPT work that has been done over the years in the process of keeping IT gear running. Script mentality was great at keeping a server or router running over time, but the data center’s $100Million dollars of assets which are constantly changing is not a server or a router. Homegrown DCIM probably feels like the next generation of SCRIPTING to some folks. (It’s not, really, it’s NOT!) IT gear, Facilities gear, everything really needs purpose-built software with support and scale and growth that can be counted on. DCIM is a strategic investment in managing the cost of doing work, not just a tactical tool.