For every story about a NetWare server running happily for four years after being accidentally secured behind drywall, you can find a dozen about the four-year-old machine that crashes eight times a day, cannot have its OS reinstalled for some esoteric reason involving unknown drivers, and that eats e-mail at random intervals. According to the stories, these are usually important e-mails, involving vital sales deals or how the company will avoid lawsuits.
Stories aside, when we wish to purchase new hardware or estimate time/budget for the coming year, we really do need a way to come to grips with the steadily rising costs associated with older hardware. We need to understand what these costs are, where they come from, and how they might be avoided.
What are the costs?
Most computers, as individual units, exist somewhere between the two extremes described above. They go though a brief period of problems early in their careers, and then settle into a long, slow process of decay. Over time various mechanical components break and occasionally the power supply gives out. After two to three years (or longer, depending on the environment) they may start to show signs of heat stress, eventually failing because they simply can no longer function to spec. Batteries fail, causing computers to crash unexpectedly or not retain hardware information after a reboot. Replacing the damaged component can cause the problem to disappear, at least for a while.
Unfortunately the costs associated with this slow decay are not confined to the hardware cost. They can also include increased support costs, opportunity/productivity costs, and even increased costs for new projects interacting with the older hardware.
Unless dealing with a machine out of legend, a single piece of hardware will rarely impact the overall budget of an entire department. However, the slow increase in hardware-related support tickets from an aging computer population will begin to take a toll on even a robust budget. The exact proportion of hardware tickets to user support/software tickets varies from environment to environment. However, when the absolute number of hardware tickets begins to increase over a long period of time it begins to tax the support organization's available time. They spend more and more time working to resolve issues that the help desk can't address, and less time doing proactive maintenance.
For example, one training company discovered that over a space of three years its hardware trouble tickets went from 10 percent of tickets in the first year to over 40 percent in the fourth, with a 10-percent increase per year. Since it did not increase its staff appreciably during that time, any noncritical task (like capacity planning) fell though the cracks as the support staff fought to keep the hardware operational. Other organizations may well have hired support staff to deal with the "increased" ticket load without realizing they were dealing with what was fundamentally a hardware issue.
Opportunity/productivity costs are even more difficult to quantify, although we generally recognize them when they occur. We know that a sales staff that cannot submit orders because their laptops cannot dial in anymore is not being productive. We recognize that if a key part of the production chain cannot communicate with others then they will be less effective. The exact costs of these issues will depend on the type of business, the beliefs of the people involved, and the practicality of actually getting work done during that time. For example, one company once estimated that its "opportunity cost" for a four-minute bootup time as opposed to two minutes was over $28 million.
Many project managers, especially those focused on software development, treat the cost of installing new products on older hardware as "extraordinary" or "unexpected" costs. The cost may simply amount to replacement costs associated with bringing a part of the company "up to spec" for a new piece of software. More commonly, or at least more accurately, the cost includes holding an entire project team in a relatively nonproductive mode while testing, implementing new hardware, programming interfaces to be legible on older computers, and similarly exciting activities. Even if these activities are planned up front they may cost more in the long run (depending on the size and depth of the problems encountered) than replacing every piece of hardware in the enterprise.
Where do we look for costs?
Generally the cost of old hardware does not come from any one source. Instead it accumulates over time, a bit from this pot, a bit from that pot, until it hangs around like an anchor around our necks. We don't notice the gradual accumulation because we are looking in the wrong places. We look at our budgets and project loads, rather than at the basic accrual of time.
Old hardware rarely demands much time from any one individual, so it slips under our radar. For example, a degraded memory chip may take four or five phone calls to support about various "unrelated" problems before it is diagnosed, and another two days to two weeks (depending on hardware purchasing requirements) to address. In this case the one chip absorbs time from the user, several support personnel, the management chain responsible for purchasing, and eventually the person who has to install the chip.
The key to finding and identifying this time lies in analyzing the support tickets with an eye towards identifying equipment patterns. Rather than looking for problem users, compare the problem incidence with hardware distributions. For example, one manufacturer that did this analysis discovered that a sudden increase in its support time requirements did not stem (despite preconceptions) from a new software distribution, but rather from a batch of 500 laptops distributed and redistributed over the course of a three-year period.
How can we avoid them?
Avoiding the costs of older hardware generally involves one of three approaches:
- Innovations designed to cope with environmental factors
- A carefully planned exit strategy
- A willingness to perform occasional OS refreshes
Take the example of one large warehousing organization. The IT department's staff members noticed that their hardware support costs increased steadily after the hardware was deployed for two years. After three refreshes in six years, they sat down and did a careful analysis of their problems. They found that the average temperature in the warehouses around the computers was 86 degrees. They also found that, although they diligently policed their systems for illegitimate software, the registries on most of the systems steadily degraded, along with the stability of the .dlls, and the system as a whole.
To address these problems they began to purchase additional cooling fans for their hardware up front, broke the hardware pool into three "groups" on different refresh cycles, and on every hardware refresh, reformatted every computer. Finally, the constant refresh cycle forced them to streamline their return process, avoiding any additional costs on their leases. After running with this program for four years, they reassessed, discovering that they cut their support tickets by 40 percent rather than the steady growth they were accustomed to. This freed up additional time for them to use in proactive activities, further reducing their overall time investment in support.
Understanding the costs associated with old hardware, where they come from, and how they might be avoided can help you when preparing your budget for the coming year.