Saturday, March 7, 2009

Key Formulas for Data Center Meaningful Reports

Written by Rakesh Dogra

Today’s data center manager is being asked to do more with less. Included on their list is the preparation of reports on efficiency, shortcomings and strengths of the data center. Creation of such reports is only relevant if provided with meaningful information based on metrics, benchmarks and formulas.

Data centers come in all shapes and sizes; therefore it is difficult to provide a specific template which works equally for all. Key metrics, benchmarks and formulas are important to start with. The focus of our discussion will be on formulas. These formulas could be strictly mathematical formulas in the exact sense of the term, or simply tips to create meaningful reports in the broad sense.

Formula 1: Since data centers are huge consumers of energy, it is important that this energy be utilized efficiently. To examine energy efficiency there are some metrics with formulas that are available.

Groups such as the Uptime Institute have been preaching the importance of corporate average datacenter efficiency or CADE. CADE is a set of four metrics which together rate the business performance of a single data center or the weighted average performance of a group of data centers.

CADE separately identifies the IT and facility efficiency of a data center, examining both energy efficiency and capital utilization. The four components of CADE with its formulas are:

• Facility Asset Utilization
- How much of a facility's power and cooling capacity is being used?
- Basic measurement concept: [Current IT load] / [Maximum IT load capacity]
• Facility Energy Efficiency
- How much of a facility's total incoming energy ends up being consumed by IT equipment?
- Basic measurement concept: [Current IT load] / [Current total facility energy], equivalent to the Green Grid's DCiE or 1/PUE under certain conditions
• IT Asset Utilization - How much IT compute asset capacity is being utilized?
- Basic measurement concept: [Average volume server CPU utilization] • IT Energy Efficiency
- How effectively does the data center's IT equipment transform energy into "useful IT work?"
- Basic measurement concept: [useful IT work] / [IT watts]. Since industry-wide definitions of useful IT work are still under development, CADE uses an arbitrary baseline value of 5% for IT Energy Efficiency.

More recently the Green Grid has developed two new metrics for energy efficiency. DCiE stands for Data Center infrastructure Efficiency and it gives a measure of the efficiency of the data center by taking the ratio of the energy consumed by the IT segment of the data center to the overall power consumption of the data center which includes other sources of power consumption such as cooling paraphernalia and so forth.

Formula 2: Another Green Grid creation and related parameter to DCiE is the PUE or the Power Usage Effectiveness and mathematically it is the inverse of the DCiE which means that it gives a measure of the total data center power consumption divided by the power consumed by the IT infrastructure, and is expected to be at least having a value of 2.

Formula 3: The power consumption formula must be understood by the data center manager if the first two formula have to be applied. Basically power consumption is measured in terms of KWh pronounced as Kilo watt hours.

If a device or equipment has a rated power of W Watts and that equipment is used for say n hours then the number of KWh consumed is given by

W * n/1000 KWh

Formula 4: Another formula which is useful to find out the heat conversion of electronic devices is given by

Heat generated = Wattage * 3.412 which gives the heat in BTU/Hr or British Thermal Units per Hour

Though all energy input to a devices does not get converted to heat, yet this formula gives a broad idea of the generated heat if cooling requirements have to be calculated. In terms of air conditioning terminology 1 ton of air conditioning is equivalent to 12, 000 BTU per hour. Of course the exact requirement could be estimated by an HVAC engineer but this formula would be good enough to give a rough estimate to the data center manager.

Formula 5: as a general rule of the thumb rather than exact formula, it has been estimated that the average power requirement for every one square foot of data center space is roughly two watts. This formula can be used to calculate the lightning load of the data center that needs to be lit up.

Another rule of the thumb assumes that nearly half of the power requirement of the data center is for cooling purposes, while nearly 36% is for critical loads. The remaining is shared by lightning (3%) and battery chagrining and UPS consumption (11%). Of course it must be kept in mind that these are just average figures and could vary substantially per situation.

Finally here are several others in no particular order.

  1. Asset Efficiency (AE) = (IT Energy Efficiency) x (IT utilization)
  2. Data Center Density (DCD) = (Total CPU Cycles) / (Total Data Center Square Footage)
  3. Data Center Productivity (DCP) = (Useful computing work) / (Total Facility Power)
  4. Deployed Hardware Utilization Efficiency (DH-UE) = (Minimum Number of Servers Required for Peak Load) / (Total Number of Servers Deployed)
  5. Deployed Hardware Utilization Ration (DH-UR) = (Number of Servers Running Live Applications) / (Total Number of Servers Actually Deployed)
  6. Facility Efficiency (FE) = (Facility Energy Efficiency) x (Facility Utilization)
  7. Storage Density = (Storage Utilization) / (Total Data Center Square Footage)
  8. Storage Utilization = (Server, Network and Backup Storage in Use) / (Total Storage Available)
  9. Storage Automation = (Human Operators) / (Storage Density)
Hence we see that there are several formulas that are available to help measure your efficiency and your data center productivity. There are others that cover areas including networking, real estate, IT production and more. We will cover them in future articles.

0 comments:

Recent Posts