When server comparison-shopping the answer may surprise you.
Many executives purchase equipment based on initial hardware purchase outlays rather than the total cost of ownership (TCO). This can be a fatal financial flaw. The premise of this blog may seem a strange topic for executives of organizations with annual revenues of less than $20 million to consider but it is a relevant, cautionary tale. Executives and IT staff frequently make decisions based upon the initial purchase price and assume that since these costs are low they have made the most economical choice. However, in the world of IT that is not always the case.
An IT Use Case
A typical use case is a mid-sized enterprise that has a new application to implement and decides to run the application on its own x86 server. After all, the Linux or Microsoft Windows-based server only costs a few thousand dollars and if more horsepower is needed, another box can be acquired. So the firm invests $5,000 on the initial hardware and another $5,000 for non-application software.
Once the server is deployed and operational, the system usage grows and additional servers are needed. Soon the firm is adding servers every few weeks or months. Before they know it, 100 x86 servers are located on siteeach one costing $10,000 (excluding the application software) to acquire. Total out of pocket acquisition expense is $1 million plus the application software costs. On top of that each one consumes energy, occupies space, demands its own storage subsystem, and requires a system administrator to run it. When these costs are added in, the TCO ends up in the $3 million range, as there are no savings due to scale.
Hidden IT Realities
Had the enterprise decided to go with a mainframe, the initial expenses may or may not have exceeded $100,000 (entry price is $75,000) but growth costs would have been far less and the TCO would be less than that of the x86 server costs. Is this scenario real? Yes, it happens all the timewith these choices and other hardware purchases. There are four reasons for these mistakes:
1. People are lulled into believing a lower cost entry price will result in a lower total cost
2. People do not project lifetime or five-year timeline fully loaded expenses well or at all
3. People think all systems are basically alike, with similar cost structures
4. People fail to properly predict the actual capacity and performance requirements needed over a reasonable time frame
Entry prices do not predict overall costs. Many vendors do all they can to lure users into buying their equipment through low acquisition costs and then lock them in with high maintenance and support costs. Moreover, in the above example, no economies of scale were realized when adding additional servers (i.e., 100 x86 servers are 100 times the cost of a single server). On the other hand, mainframe costs decline proportionally as capacity increases (e.g., twice the capacity may run 1.8 times the initial fees or six times the capacity could run 1.5 the entry price.