Are you worried about total ownership costs (TOC)?
Don't place your faith only on changing technology.

Replacing old PCs with NCs or switching operating systems is not the miraculous answer to the rising costs of computer networks. You'll save more money if you improve reliability and security, and become a superb manager of the technology you now have.

Until recently we rarely heard much about the excessive costs of personal computers, even though they can amount to as much as 40% of the salaries paid to their users. Computer people kept quiet, because they were preoccupied with obtaining money to buy hardware and software.

Top management only began to inquire about the full life-cycle cost of computing when they recognized that computer ownership increasingly includes costs that show up only in the users' operations. Today, user-incurred costs such as "futzing" (wasting time on PCs), improvised learning, assistance from fellow employees and end-user attempts to become computer experts are showing up in the TOC estimates published by consultants and vendors.

Downtime Estimate

Frequently, these disruptions can lower employee productivity and hurt revenues. In business-critical applications, computer downtime equals the costs of idle labor. I have just completed a study of a network with 2,500 clients, 80 servers, 50 bridge/routers, 200 hubs and 160 printers. The unplanned downtime for the clients was 1.6 hours per month per person. When I added the interruptions caused by the servers, routers and hubs, it added up to 2.8 hours. For the entire network that was 84,000 hours of dysfunctional time a year. With user compensation averaging $32,000 per year, plus benefits, that could add up to a large waste of resources.

The Total Cost Of Ownership Model

Interpose, based in Altamonte Springs, Fla., offers the most comprehensive software for estimating total ownership costs, including expenses that do not show up in the IT budget. In addition to calculating hardware and software component costs, it also includes allowances for all other IT expenses:

• Network management (such as troubleshooting and repair, performance tuning and user administration);
• Systems management (such as technology evaluation, licensing, asset management, network security and installation of upgrades);
• Storage management (such as storage capacity planning, backup and archiving, and disaster planning);
• User support (such as training, vendor liaison, equipment relocation, trouble diagnosis and help desk operations).

• Lost productivity and lost revenue from downtime. (The model estimates these losses from data, supplied by users, on factors such as salary, what portion of the person's time is critical, planned downtime, unplanned downtime, what portion of the population is affected, and how mission critical are those people.)
• Time spent in formal training;
• Cooperative training, when the user's peers interrupt work and offer tutoring;
• Casual training, such as learning from trial and error, or by perusing a manual;
• The cost of correcting errors resulting from the inability to operate the system;
• The ``futz factor'' -- the costs of users spending work hours using their PCs for entertainment and private purposes.

The Downtime Effect

Other end-user inefficiencies would account for only 12% of the TOC. Futzing amounts to only 2.9% of the TOC. These results intrigued me. The key influences on TOC were not the standard cost elements that everybody talks about, but network reliability.

Next I wanted to assess the value of the work that wasn't done because of downtime. To estimate that value I related it to the salary: the higher the user's salary, the higher their potential impact on loss in revenues. In this way I assumed that the higher the compensation the more likely is that personŐs effect on business outcomes. Salary will be also a measure of lost productivity in case that person is unable to function because the essential computer support has collapsed.

I also examined the total costs of ownership for the identical 2,500 client network, assuming that it supported highly paid users (earning $90,000 a year). The losses due to downtime were even higher. For highly paid analysts, using low reliability network PCs (where downtime was 3.2 hours a month), the average cost per client was $29,100. In that case the costs of hardware and software would amount only to 8% of TOC! If these analysts were currency traders, and totally dependent on workstation up-time to conduct their business, the TOC would be much higher and the importance of hardware and software is not significant.

The following table illustrates the sensitivity of TOC to downtime and employee compensation deploying the identical hardware and software:

For 2,500 Win95 clients, 80 WinNT servers, 50 bridge/routers, 200 hubs and 160 printers the TOC can range anywhere from $29,100 to $8,503 depending on the value of the people sitting behind the displays and the business utility of the network. The choice of hardware and software technologies, that everyone talks about, can amount to anywhere from 8% to 27% of the TOC. Clearly, it is not the choice of technologies but the management of people and networks that are the decisive influence.

CIO Implications

It's not technology, but management effectiveness that drives the total cost to an enterprise for using networked computers in the execution of critical business tasks. It is management that can exercise the greatest influence on how an organization overcomes the revenue impact when networks fail to function. Only effective network management can deliver reliability and responsiveness to customer needs. Identical servers supporting identical clients, using identical routers, identical hubs and identical printers will show widely different TOCs depending on who uses them, what is the importance of computers in supporting critical applications and the reliability of the networked services.

Top management is finally starting to ask questions why overhead costs are up despite the introduction of supposedly inexpensive computer networks. This time around it will not be prudent to offer yet another technological fix as the cure for bloated IT budgets and rising overhead costs.