RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2
2015/08/07 11:35:28

DPC Measurement of profitability

In 2010 the industry organizations and government departments of the USA, Europe and Japan came to the basic agreement about a method of calculation of energy efficiency of data processing centers. Let's consider the approved metrics in more detail.

The directory of TAdviser of Data centers of Russia and technologies for data centers.

Content

This agreement approves a uniform metrics for expression of energy efficiency of DPC of different types worldwide. As a result there was a standard criterion which can be applied both to assessment of profitability of data processing centers, and to determination of efficiency of methods of economy of energy in other structures.[1]

The reached agreement is unprecedented on extent of the international cooperation. Process of signing of the contract took place under control of the American industry consortium Green Grid, and the U.S. Department of Energy, the U.S. Environmental Protection Agency, the European initiative of drawing up the set of rules for DPC and the Ministry of Economics of Japan supported the agreement.

As a preferable measure of energy efficiency agreement parties approved an indicator of Power Usage Effectiveness (PUE) — effectiveness ratio of use of energy. PUE developed in Green Grid sets a ratio of total amount of the electric power consumed by DPC, and the amount of energy spent for power supply of the IT equipment. The result shows how many energy it is lost on mechanical and electrical systems.

The choice of PUE is not accidental. This indicator is most popular for assessment of profitability of data processing centers, and the large companies led by Microsoft and Google, publishing data on the new DPCs to characterize their energy efficiency, PUE use.

However the standard method of calculation of PUE does not exist yet so a lot of work still is necessary to the organizations participating in the agreement. In particular, it will be necessary to regulate a method of measurement of the total quantity of energy consumed by a data processing center. Within the agreement it is also going to develop additional indicators of assessment of energy efficiency.

"The expanded working group into which representatives of each of the organizations participating in the agreement will enter will continue this initiative and later this year will hold a meeting for assessment reached", said in the statement of Green Grid.

The agreement is signed according to the results of a number of the meetings held by concerned parties. In January representatives of the U.S. Environmental Protection Agency met members of the organizations of the industry of DPC — 7x24 Exchange and the American society of engineers on heat engineering, refrigerating appliances and conditioners of air. Then in February the international meeting at which practices of these organizations were discussed took place.

Use of energy in DPC becomes subject to more and more close attention. Because of increase in prices for the electric power and increases in number of the powerful IT equipment data processing centers became for the top management of the companies of "painful points" in article of monetary costs in recent years. The enterprises are also forced to treat more attentively expenditure of the electric power in DPC because of ecological restrictions, in particular because of the new rules introduced in Europe on regulation of emissions of carbon in the atmosphere. Similar rules, most likely, will be adopted sooner or later also in the USA.

The consortium Green Grid which activity is directed to development of standards in the field of energy efficiency of data centers offered three new metrics in March, 2011. It is expected that in the future these metrics will be used on an equal basis with the PUE standard (Power usage effectiveness is efficiency of use of the electric power), which is already used in the industry. Metrics are intended to assist operators in planning, design and operation of data centers.

The first metrics – DCcE (Data Center Compute Efficiency – Efficiency of calculations in data center). This metrics will allow owners of data centers to estimate efficiency rate of use of computing resources. This metrics is not connected with performance of computer systems, though is held for use together with like those.

The second metrics, ERE (Energy Reuse Effectiveness is Efficiency of reuse of energy), will allow to measure efficiency of use of energy (first of all, thermal) which is created in use data center. Unlike the existing metrics of PUE where reuse of energy within data center is already considered, ERE is directed to measurement of energy which is used out of data center.

The third metrics, WUE (Water Usage Effectiveness is Efficiency of use of water resources) is intended especially for application in the data centers using water cooling. The metrics includes two mathematical models: standard WUE and WUE (source). In a latest model, in addition to the water consumed in a point of placement of data center all volume of the water extracted from a source for the power generation consumed in data center is considered.

Some recommendations on the way to more effective DPC

Expense control over ensuring functioning of data center can significantly affect expense structure of the company. The main directions — transition from physical servers to virtual and upgrade of the server hardware on energy efficient. In both cases investments are required, but return from them can be notable.

Here it is possible to update licenses upon purchase of new servers that will undoubtedly affect to the best. Isolation improvement from each other of hot and cold air flows indoors, use of thermostats really reduces the cost of the consumed electric power.

So, several councils for more effective work of DPC.

Virtualize everything that you can

Virtualization allows to reduce the volume of the used space, to use more economic cooling systems and more effective management.

Consider the restrictions inherent in virtualization

It is possible not always if a system services thousands of users or some vertical applications.

Consider acquisition of licenses

Complex agreements can extend to virtual environments.

Analyze a situation with energy consumption in different time of day

Possibly, it will help to optimize workloads on servers.

Consider placement of the heat-generating equipment

Instead of simple alternation of cold and hot passes, apply more thoughtful approach.

Consider the possibility of use a blade servers

Analyze the cost of upgrade and possible savings in process of energy consumption, the areas and service.

Control the number of physical ports

The large number of physical connections to one server leads to complication of network infrastructure.

Consider hardware upgrade in use

Modern servers allow higher values of ambient temperature that allows to save when using cooling systems of premises.

See Also