The CIO's Guide To The Evolving Datacentre

A CIO’s life in 2013 can be fairly daunting. Looking at the way trends have been picking up over the last decade, some aspects of technology management seem to have gotten easier and more distributed, but there is a certain sense of complexity that has crept in – especially at the datacentre level.

The last few years saw an upswing in distributed services, and while cloud was still a high level concern, it has more or less become an imperative today. With an evolution in the way enterprise systems are designed and implemented, CIOs need to start looking at things a little differently.

One of the sessions at TechEd 2013 highlighted this very change in implementation methodology, design and management. It’s quite obvious that today, given the entire enterprise technology ecosystem, datacentres are getting bigger and more distributed. While the actual number of physical servers in a datacentre may have reduced slightly over the last few years, there has certainly been a spike in the number of virtual machines, the amount of cooling required for high-end servers, networks and not to mention rising land costs.

More from Biztech

Future Group - Reliance Retail Deal approved by CCI RBI ban on cryptocurrencies takes effect; prohibition could force investors to tap the black market

These conditions are significant roadblocks and a CIO can have a tough enough time trying to keep operational costs at a minimum.

As all of you know, the last few decades have been fairly disruptive in terms of the technology that CIOs have access to. By taking control of the datacentre and reducing operational costs by initiatives like virtualisation, CIOs have been able to somewhat consolidate their datacentre environment. The big question here is that while reducing operational costs has been fairly simpler over the last few years (through high impact initiatives like virtualisation and so on), reducing your operational costs today can be a much more rigorous exercise – one whose fruits might only appear after a good amount of labour.

Senthil Kumar G, Partner Enterprise Architect, Microsoft Corporation took up a session on some of the things that Microsoft has done to make its datacentre more efficient and not to mention more cost effective to run.

“Over a period of two decades, Microsoft has been able to reduce its PUE rating from about 2.2 to the range of 1.8 to 2.0. Now, this might seem at first to be a minor shift, but it is a ratio and a representation of how much of a change has been made, and in that regard it is an excellent statistic,” informs Senthil.

CIOs need to remember that the datacentre game has now become about the little things. In his session, Senthil spoke about the little things that can have large impact on the datacentre. Try and look at it this way – you may be in possession of X number of servers that run a number of virtual machines in each channel. You have your network switches that are constantly transmitting data especially given the fact that you, as a CIO, may choose to distribute your enterprise services over the cloud, which is something that Microsoft recommends CIOs start doing immediately. And, on top of that, you have associated devices that manage and track these servers and all this enterprise hardware.

So at the end of the day, it’s not really about how much hardware you have in there, it’s also about the amount of equipment you have provisioned to manage your hardware. It’s about the UPS and the backup systems you have deployed, the heat exchangers and cooling units that you are using. Cooling, in fact, is one of the biggest pain-points in terms of power consumption in datacentres.

So given all these devices, CIOs need to really differentiate how they are planning and building their datacentres. According to Senthil, it is important to be able to learn some of the best practices from technology providers themselves, who are now really building strategy into their technology practices.

Just imagine for a second the density of hardware in the Office 365 cloud, or the Azure cloud. “You have a really large number of machines (physical and virtual) that form the cloud that provides Microsoft’s first class services to its user base – managing that is not easy and our datacentre teams really work hard to keep things efficient,” explains Senthil.

The important thing to take away here is that CIOs need to really build this operational excellence into their team right from the start. CIOs really need to measure power at every single point and figure out an optimisation strategy that can tie into a broader datacentre strategy. Some of the things Microsoft recommends in its experience are to keep a tab on the tertiary or supporting devices that keep the rest of the infrastructure running optimally. If your pain point is cooling, look at more efficient ways of cooling. Find other ways to cool you datacentre, or redesign your facility in a way that you are cooling more devices with fewer units.

“It is the little things like this that CIOs must get their teams to focus on, if they really want to optimise on datacentre costs and efficiency especially in today’s dynamic enterprise environment,” signs off Senthil.