All organizations regularly engage in strategies for server consolidation and capacity planning. The goal is to eliminate multiple individual servers and maximize available resources to adequately support business expansion while reducing CAPEX and OPEX.
Despite the importance of this repeating initiative, and despite the fact that consolidation and capacity planning has been performed many times over, I observe in many companies it is more of a black art than the simple and thorough process it should be.
From my experience, there are five basic, but important, steps to follow and I have mapped them out.
Define Hosts/Servers for Consolidation Analysis
This is a pretty obvious first step, but what’s often missing is a more thorough analysis of the current environment and costs associated – as well as the requirements of business strategies. Thinking about CAPEX and well as OPEX, do more than just a census or a “back of the envelope” plan with your hardware vendor.
Think holistically. Think about leases and leasing or co-hosting. Consider the age of your current servers and problems you may be having in supporting them or staffing to keep them operational. Look at existing floor space requirements and costs associated with that as well as power constraints or other needs.
Define Time-Frame (Baseline) to Use – and Relate to Business Peaks
For IT to be a good business partner, it seems obvious to me that you need to monitor your data center performance 24x7x365. If you’re not doing that, you have nothing to accurately base your consolidation or expansion decisions. You should be looking at business peaks and IT peaks and mapping out the correlation of those two.
If you have been using a performance monitoring tool like Galileo Performance Explorer, you have accurate information to base sound decisions on. If you do not have that data, consider installing monitoring software and gather some data for the next 30 or 60 days to give you a baseline.
It would be ideal if this period included business peaks and valleys, but if not use the baseline data to map to the business cycle to come up with some algorithms of what is required. (I will address this further later in this post.)
Determine CPU, Memory, Network and Disk Usage (for Step 2)
This is a fairly straight forward project if you have done your homework in step 2. A consideration here is if there is any move from physical to virtual environments. The move is not a one-to-one correlation as you need to account for extrapolation of network packets.
Define Growth Projection
There are two ways to do this:
- First (and best) is to get this information from the business plan. It is imperative your IT is closely aligned with the business – can’t stress this enough. The business world today relies on IT more than ever, and for CIO’s and IT administrators to be part of the solution, IT must have steady and constant communication with business units. Granted, if you did the performance monitoring required in step 2 you have a good perspective of what happened in the past, but that is not necessarily a predictor of the future. Successful server consolidation and capacity planning requires business alignment.
- If there is not enough information available on past IT performance, the second option is an algorithm I’ve used in the past that works well.
- First, use your Baseline and increase by 25%. That gives you a base for the coming year.
- Then, increase by 15% year-to-year over the next 3-5 years.
Is this as accurate at the first approach? No, but if you’re missing historical performance data, you need to make some assumptions and this method has worked for companies in the past.
Recommend Consolidation and Capacity Solutions
Your consolidation plan will now be based on a realistic assessment of the required capacity for support of the business plans. It will be built to support the required availability, footprint and performance requirements while efficiently mapping out location, energy, and resource costs and optimizing your CAPEX and OPEX.
Feel free to share your experiences/comments below. And if you’d like to walk through any server consolidation best practices for your specific environment, please contact me.