(This is the second article in a five-part series.)

Ask any expert: When virtualization rollouts fail, poor planning is usually to blame. “A lot of people just purchase equipment without knowing the requirements,” says Matt Cavanagh, president of Flytrap Technologies LLC, a data center integrator in Tolland, Conn. The result, he adds, can be budget overruns, listless performance, and schedule delays. “Planning is everything,” Cavanagh cautions.

Before executing a virtualization deployment, be sure to complete these preparatory steps:

1. Map out a long-term plan: Even if you’re planning to virtualize only part of your infrastructure initially, create a two- to three-year virtualization roadmap before buying or deploying anything. Otherwise, you’re likely to make your software and hardware purchases piecemeal and end up missing out on money-saving volume licensing deals.

2. Calculate your consolidation ratio: A “consolidation ratio” is the number of virtual servers you place on each of your physical hosts. Many companies arrive at that figure through trial and painful error by simply stacking virtual machines on top of each other until performance slows to a crawl. “It’s sort of capacity planning by bloody nose,” observes Bob Gill, managing director of server research for analyst firm TheInfoPro Inc. of New York, N.Y.

Executing a rigorous capacity planning assessment is a better approach. Such studies utilize hard data on how much processing power, memory, disk space, and network I/O your infrastructure requires. PlateSpin Ltd., CiRBA Inc., and VMware Inc. are three of many companies with capacity planning products that gather such statistics automatically. For maximum accuracy, collect at least 30 days’ worth of data. Your capacity planning tool will then compare those figures to your host servers’ performance specs and recommend a suitable consolidation ratio.