Calculating TCO for Energy
Realistically estimate electric use and cost to compare servers
IT organizations and the consultants and salespeople who serve them often generate business cases to project costs of alternatives. Though commonly termed total cost of ownership (TCO), these analyses involve variable costs of IT infrastructure. A trend of the exhaustion of data-center electrical capacity has raised the importance of electrical consumption and costs. In this article, you’ll learn how to estimate server electric consumption and calculate the resulting operating expenses for business cases.
Server Electric Values
The first information needed is the amount of electricity each server consumes. This might not seem straightforward. A 70-watt (70 W) light bulb consumes 70 W when on. A 1725 W-rated server doesn’t consume 1725 W, though. Even for the same brand and model, different configurations consume different amounts of electricity.
This value is the nameplate rating. The 1725 W example is used by the system central electronic complex enclosures of many Power Systems* servers. Unlike a light bulb’s value, the nameplate rating doesn’t state the server’s electrical consumption. It states the power supplies’ capacity. Calculating electrical use or cost by the nameplate rating results in significant overestimation.
Finding Typical Power Values
What should be used to calculate server electric use and costs? Use typical power, the estimate of a server model/configuration’s average electric use. Because modern servers often have large variations in configurations, published typical power values are rare. To get reasonable estimates of typical power for servers built since about 2005, use the vendor’s power calculators. Different brands often have separate power estimation (see “Power Sources”).
For servers built before 2005 (e.g., eServer* pSeries* and RS/6000* platforms), find the site planning documentation. A copy of the Site and Hardware Planning Information, document SA38-0508-12, is available at http://bit.ly/o5h5Hn. Many of the models listed have typical power values.
Site-planning documentation is the authoritative source for electric and space information for all vendors, brands and models. Always use site or installation documents to get values when a calculator isn’t available. Many older x86 servers will have electrical information in either an installation guide or user guide. Avoid brochures, datasheets and at-a-glance type marketing materials.
Estimating Typical Power
If the vendor doesn’t have a calculator and the site-planning information doesn’t list typical power, estimate based on either the category or derivation based on the nameplate rating. Categorically, commodity x86 servers can be estimated reasonably. Average typical power consumption for servers ranges in the following categories:
- 1U rackmount x86: 300 W-350 W
- 2U rackmount, 2 socket x86: 350 W-400 W
- 4U rackmount, 4 socket x86: average 600 W, heavy configurations 1000 W
- Blades: average chassis uses 4500 W; divide by number of blades per chassis (IBM BladeCenter* H is 14 per chassis, so 320 per blade server)
To estimate within these ranges, consider that electrical consumption increases with higher clock-speed CPUs, larger numbers of memory cards such as DIMMs and physical disks, and with greater processor utilization.
Servers that don’t fall into one of the listed categories can have typical power estimated by multiplying the nameplate rating by 70 percent. This estimation is reasonable only for a large population of servers, such as a whole data center. It’s not accurate with any granularity, and certainly not at the single-server level.
If the example server is unplugged and its workload deployed elsewhere, $333 per year in electrical expense will be saved.
comments powered by