The computer room of a power grid company (called PGC for this article) was bursting at the seams. With around 1,200 virtual environments running on 262 physical servers, the space teemed with boxes, racks, and cables. And each item that squeezed into the room took time and money to maintain.
When a new server technology was introduced that would enable the data center to use fewer servers, the company’s IT leaders were eager to initiate a server consolidation project. There was, however, a fundamental question they needed to answer. “If we consolidate the virtual environments to the new technology, how many servers would we need?”
Digging into the Data
Without that data, the project would have been delayed and more difficult to execute. Even tracking a month’s worth of server usage today would not help—that’s because it was not their busy season. Peak energy usage came in the dog days of summer and correlated with server usage. To accommodate the high points of demand, they needed to know what happened in August.
The data allowed them to establish baseline needs and the capacity of each of their servers. While some were out of capacity, at 90 to 95 percent utilization, others were only 10 percent utilized.
Luckily, PGC had been monitoring their server usage using Galileo Performance Explorer®, an infrastructure performance management tool, for a couple of years. So I reviewed the historical data. Not satisfied with August’s monthly average usage, I looked at the busiest days of that month. Nothing less would enable them to keep operations running smoothly as air conditioners worked overtime during future heat waves. The data allowed them to establish baseline needs and the capacity of each of their servers. While some were out of capacity, at 90 to 95 percent utilization, others were only 10 percent utilized.
Fun and Games
Within Galileo, I created virtual views of the new servers and played a “what if” game to see how many of the new servers they needed. Since they anticipated 15 percent a year sales growth, I added the same percentage to their server demand projections for the next three years. And, just to be extra conservative, I added a 20 percent buffer on top.
The answer? They needed 28 servers.
Interestingly, when PGC had also asked the technology vendor how many servers they needed, the vendor had told them “60.”
Assumptions Are an Unaffordable Luxury
…their technology vendor had assumed all the existing 262 servers were maxed out….this assumption ensured….an exorbitantly expensive solution.
Why was there such a big difference in the two answers? Because their technology vendor had assumed all the existing 262 servers were maxed out. Since this assumption ensured that the customer would not run out of capacity, it was safe. Sadly, it was also an exorbitantly expensive solution.
As today’s IT budgets inch up slowly and the demand for resources climbs more rapidly, IT leaders are challenged to do more with less. In this environment, untested assumptions are a luxury they can no longer afford. To make the most efficient use of resources, they have to take a strategic approach to server consolidation.
Clearly, if you don’t know what capacity you’re using today, you cannot plan effectively for the future. On the other hand, making decisions based on historical data enables you to optimize your ROI on technology investments. You not only save on the initial capital expenses but also on the following:
Because PGC took a data-based, analytical approach to server consolidation, the company now only needs to update firmware and care for 28 servers, instead of the 60 their vendor recommended.
Remember, when servers roll off their warranties, maintenance costs rise. Manufacturers typically design technology to have a life of 3 ½ years, and yet warranties usually expire after three years. The older your disk drives, the more likely they are to fail, so your maintenance costs can climb rapidly.
Because the company uses fewer servers, power cords and rack space, the computer room is no longer jam packed. And the bonus is that there’s room for expansion when necessary.
Servers are on 24/7/365, perpetually consuming power. If you can minimize the number of servers, you naturally lower your energy costs.
IT administrators are in demand and hard to find. The more servers you have to manage, the more technicians you need. With fewer servers, you minimize staffing requirements and the time it takes to recruit and fill positions.
Server consolidation can save you on the initial investment as well as on maintenance, space, energy, and people. To do it right, however, you cannot rely on the assumption that you’re using all your server capacity today. You must collect data on usage patterns and use it to make an informed decision that maximizes your ROI.
To learn how Galileo’s cloud-based infrastructure performance management tool can help you, schedule a demo now. Also, ask about our free server consolidation assessment.