Running a data center is really costly. From monitoring the infrastructure to cooling every step that you take means some payment. Actually, there are many ways which you can “cut the fat” on your data center running costs. These actions will take some time but you will begin to see the effects in a very short time.
Let’s cut the fat from the servers. They are important for three reasons: first they use resources, such as electricity and cooling, second, you are apportioning some time of your IT team to the monitoring of these servers and you are paying for the maintenance. There are two steps you can take:
- Identify zombies: In an enterprise data center you probably have servers that are already replaced but still continuing to run. The reasons can be the fear that something may be broken because the infrastructure is not very well known or for “just-in-case” fears. If you cannot shut these servers down, virtualize them. If you fear that the replacement server may not be functioning as the old one, shut the old one down and see what happens.
- Consolidate: Since virtualization, bundled with computing power and storage looks like an infinite resource, bringing the fallacy that any virtual server costs nothing. This fallacy brings in having a lot of underutilized servers to the data center. In a data center I have seen many servers each running only one small ISS application with less than 50 hits. We have consolidated all the applications them to just two servers, which were clustered. Not only we got rid of 2 physical and a couple of virtual servers, but also we achieved high availability. We have placed each cluster node on different hosts running on different blade/SAN, achieving high availability both on the operating system level and the hardware level.
The next things to check are the maintenance agreements and cloud hosting options. To keep your data center up and running you have a maintenance agreement, most probably from the vendor. Once you have consolidated your data center, think about what you can move to the cloud. In one of my clients, I have identified that a big portion of the IT spending was on the maintenance of the DMZ servers, serving only web content. Some were serving IIS applications, some were serving PHP applications using MySQL databases, some were using the WordPress or Joomla applications. I have offered moving them to cloud hosting companies and getting rid of all the servers. Despite a few questions, my offer seemed logical and we have migrated the applications to hosting companies. Next, we have changed the obsolete servers’ maintenance agreements from 7/24 to business hours and redeployed the servers for testing and development, which the company did not have this infrastructure. I have advised the freed up cash to further consolidate and simplify/standardize the web sites and applications. Imagine how many birds we hit with such a decision. Be sure that there is a similar scenario in your data center.
Once you have went through all the pain, you can start to work on your licensing issues. After all, you may not be needing that number of licenses or that level of licensing agreements. Just think about this: if you have already moved to Office 365, why are you still paying for the Windows Server, Exchange Server, Lync Server and SharePoint Server licenses? This may not be a 100% right scenario but you get the point.
Take a look at the vendor discounts. You may have three places to save:
- Rework your vendor agreement. Try to consolidate different vendor agreements (most probably on the hardware vendors) into one and ask for discounts. Never ever try to squeeze every penny from your agreement. This will create an adversarial relationship with the vendor and will definitely backfire.
- Ask for assistance: Ask your top vendors to perform a check in your data center to see where you can save. Their experience will surely identify points here and there that will help you save. They will charge you by the consultancy and you will cut your costs and save more. A perfect win-win scenario for both parties.
- Clip some coupons. Probably have some discounts lying here and there that came with your purchases. One of my clients had discounts that came with the tape purchases from the vendor and were lying around unused. Using them did not make the company fly but who doesn’t want a couple of free tapes every year?
You may be overspending in two areas: disaster recovery and consultancy. Analyze your disaster recovery plans. Do you really need that much of everything? If you have a disaster site and an off-site to store your data backups (tapes, let’s say) do you really need an additional international disaster site? If yes, do you really need a domestic side? OK, you may need both, but reconsidering is the point here. Similarly, you may have an excessive number of consultants paid by an hourly rate. Check what they are doing, check their projects and their progress. I have seen consultants working in the company for a project for weeks for a project that has lost its business value. If the consultant’s work is so important, think about hiring a full-time employee. Consultants are at least 2 times more expensive than full time personnel.
You can start working on these items today with your IT teams and complete your projects in about a year (considering a data center with about 500 physical and virtual servers). With IT teams, I mean not only the infrastructure team, but also your developers and your desktop support techs. Not only you will get everyone on board but also you will get valuable, informed advice right from the field.