By Jerrold M. Grochow and Kelli Trosvig
Jerry Grochow is senior advisor to NET+, Internet2. Kelli Trosvig is vice president of Information Technology and CIO, University of Washington.
University IT leaders need to understand the costs to procure, develop, and operate the services their departments provide in order to evaluate maintenance, reimplementation, and replacement of those services. Determining this "total cost of ownership" (TCO) has taken on added importance as opportunities for sourcing services via the cloud present themselves, where all costs are now rolled into a single service cost. Cost comparisons between locally provided services and cloud services are difficult because some costs of providing on-campus systems are not part of the IT budget — or even tracked by the accounting system — and other costs (and benefits) may be difficult to quantify.
For example, the TCO of a locally sourced learning management system (LMS) would include hardware, software, and system operators, as well as space, cooling, and electricity (for both hardware and people), some of which are costs in the IT budget and others not. Some (such as the cost of the LMS software itself) are obvious while others have to be approximated or allocated (e.g., the cost of a database management system), if they are to be counted at all. If the LMS service was provided by a cloud vendor, all of these costs would be rolled up into a single bill from the service provider, including that service provider's profit! The costs of people to manage the system, implement upgrades, and provide support to the community are also part of the TCO, and some of these may not be tracked or may be in several different departments. For a complete TCO analysis, all costs have to be brought together (hence the concept of "total cost"). This is particularly significant since operational costs would likely be very different if the LMS were provided on campus or via the cloud.
Another concern in determining TCO are hidden subsidies, i.e., costs that are not attributed to the people who benefit, thus providing a subsidy to those people. Research computing resources provide an example of this situation. While researchers typically buy their own computing equipment, they often do not bear the cost of space to house it, power to run it, or equipment (and more power) to cool it. Most often, those costs are borne by the institution’s facilities department and rolled into institutional overhead. Even when they are charged back by the facilities department, they rarely make their way down to a particular research grant due to various government accounting rules. With the significant costs to power and cool today's research computers (as much as $60,000 per year for a single rack) versus providing lighting and cooling for more typical office or research space, researchers with large banks of computers are getting a cost subsidy over their less compute-intensive brethren — as well as over researchers who are moving their work to the cloud where no such subsidy exists. Current rules governing charges for federal indirect costs further complicate the comparison between local and cloud computing costs because equipment is exempt from federal indirect cost charges (routinely over 50% for most large research universities), whereas cloud services are not.
Another subsidy that is common with research computing has to do with non-official uses of computers purchased for a particular project. In general, the research project that purchases a computer bears the full cost whether it is fully used or not. When a particular researcher is unable to fully utilize a local computing resource, that spare capacity may be made available to other researchers at little or no charge. This may be a perfectly reasonable thing to do (unused computing cycles are lost forever — just like unsold airline seats and empty hotel rooms), but using these resources represents another hidden subsidy, albeit to different researchers than the ones purchasing the equipment. One of the selling points of cloud computing is that you only pay for what you use, but this also means an end to these types of subsidies, something that should be considered when comparing local versus cloud computing costs.
The ECAR TCO for Cloud Services Working Group1 has identified several hidden subsidies and indirect costs that affect TCO, including energy and electricity costs, facilities costs, some types of departmental labor (e.g. the graduate student who helps out with computer support), centrally borne costs such as business continuity services, and many types of infrastructure costs. Presumably, the purpose of a TCO analysis is to understand the total cost to the institution of providing a service, and not just to a particular organization’s budget. That way, the TCO analysis can be used to compare current service deployments to alternate ways of providing them (such as in the cloud). It becomes particularly important to identify costs that are not directly accounted for in the university's or IT department's accounting system (i.e., "unhide" them) to complete the analysis. While it is true that subsidies from one group to another "don't affect my budget," they are all costs to the university, and if the real TCO of a service can be reduced, the university has more resources with which to fulfill its educational and research mission.
Note
- The ECAR TCO for Cloud Services Working Group constituted earlier this year is producing a report that will detail the various costs and other considerations in determining total cost of ownership both for locally sourced and cloud-sourced services. For more information, see https://www.educause.edu/ecar/ecar-working-groups
© 2014 Jerrold M. Grochow and Kelli Trosvig. The text of this EDUCAUSE Review online article is licensed under the Creative Commons Attribution 4.0 license.