December 17, 2009 at 04:33Server cooling - “more” does not always mean “better”
A survey conducted by one of the leading manufacturers of refrigeration and air conditioning systems for data centers, by Liebert, showed that 66% of respondents maintain a temperature in data centers of no higher than 21 degrees C, and none higher than 24C.
The horizontal temperature is in US Fahrenheit.
CRAH is a Computer room air-handler, in our opinion - air conditioning.
At the same time, the recommendations of ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers), revised last year, recommend an upper limit of the working range of 27C at the entrance to the server .
This state of affairs is well confirmed by my "observed practice."
Traditionally, owners and operators of data centers are guided by the rule "the colder - the better for electronics", of course, until the dew or snow starts to fall in the hall;)
It is generally accepted that every extra degree above at least 21C makes the death of server hardware components from leaps and bounds overheating.
This is often taken for granted so much that it is not even discussed.
However, the current trend on Green IT, or on energy-saving technologies in data centers, could not pass by the question of the actual needs of server equipment in the optimal operating temperature.
The results of a study of the issue may seem rather unexpected. Everything suggests that the current situation with the temperature in data centers stably below 22 degrees is not optimal either from the point of view of equipment operation or from the point of view of the component's operating life, and the danger of high temperatures is sharply overestimated.
So, an article in The Register comes out with a catchy headline traditional for the Register: " Intel says data centers much too cold - Frozen assets a waste of cash" - "Intel claims that data centers are too cold - frozen assets spend money aimlessly."
Nevertheless, the topic is not raised at all "yellow."
The industry-wide reassessment of the danger of high temperatures for data center equipment is confirmed, for example, bya recent Intel study , in which 896 identical blade servers, were divided equally into “control” and “test” groups (8 cabinets, 4 blade chassis in each, 14 blade servers in each chassis, a total of 448 in each site ) The “control” was cooled in the traditional way with the help of a closed-circuit air conditioner, and the test was cooled by the usual “outboard” air, according to the “open” scheme, with its minimum dust removal and without humidity control.
The aim of the study was to prove the possibility of building cost-effective data centers and reduce cooling costs (it is no secret that the costs of a modern data center are very significant. The cost of powering air conditioners can be from a quarter to half in the total energy consumption)
Source:http://dashboard.imamuseum.org/node/830
Vertical - kilowatts, horizontal - days of December. Winter is in the yard, and cooling costs are apparently minimal. The graph does not apply to the Intel test and is given only as an illustration of the typical ratio in the power consumption of a data center.
Also:
Source: http://www1.eere.energy.gov/femp/program/dc_energy_consumption.html
In the case of an Intel experiment, despite the temperature fluctuations in such a "non-conditioned" data center, sometimes rising to 32 degrees, the failure rate compared to the control group differed a little more than half a percent (4.46% versus 3.83% for the "traditional" Intel data centers on average, and 2.45% for the control group, which, in general, is within the "statistical spread").
An even more interesting situation is with the temperature dependence of failures for hard drives. For example, in 2007, a report was published by Google engineers who were studying the frequency and causes of hard drive failures in the server centers of their company (data from about 100 thousand drives were processed, and the study lasted about nine months).
One of the interesting results of that study indirectly confirms the recommendation of ASHRAE, regarding the temperature regime in data centers. So, according to the observations of Google researchers, the probability of failure of hard drives increased sharply when their temperature dropped below 30 degrees, and the lowest probability of failure for the observed group of drives corresponded to a temperature of as much as 40C!
At 40 degrees of operating temperature (all measurements were carried out using SMART sensors), the probability of failure did not exceed 1% AFR (Annual Failure Rate, annual number of failures), an increase to 50C doubled the AFR to 2% (higher temperatures were not observed in the data center) .
But lowering the temperature to 20C paradoxically increased the probability of failures by almost ten times, up to 10% AFR!
On the graph, the histogram bars show the relative number of disks having a particular temperature, points with t-shaped strokes — AFR values for a given temperature and its statistical spread, increasing with decreasing number of “participant” disks with a given temperature.
A noticeable increase in the number of failures at elevated (significantly!) Temperature was observed only for disks with an age of more than three years.
Conclusions:
It is entirely possible that the “the colder the better” approach has outlived itself. This paradoxical, at first glance, conclusion is confirmed by some statistical results, which suggest that, perhaps, we underestimate the “elasticity” of the temperature regime of modern equipment, and its ability to tolerate “elevated” (in our opinion) working temperatures.
In addition, every degree by which the temperature in the data center can be raised there is a direct savings in electricity bills.
First published on the blog http://proitclub.ru/ .
The horizontal temperature is in US Fahrenheit.
CRAH is a Computer room air-handler, in our opinion - air conditioning.
At the same time, the recommendations of ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers), revised last year, recommend an upper limit of the working range of 27C at the entrance to the server .
This state of affairs is well confirmed by my "observed practice."
Traditionally, owners and operators of data centers are guided by the rule "the colder - the better for electronics", of course, until the dew or snow starts to fall in the hall;)
It is generally accepted that every extra degree above at least 21C makes the death of server hardware components from leaps and bounds overheating.
This is often taken for granted so much that it is not even discussed.
However, the current trend on Green IT, or on energy-saving technologies in data centers, could not pass by the question of the actual needs of server equipment in the optimal operating temperature.
The results of a study of the issue may seem rather unexpected. Everything suggests that the current situation with the temperature in data centers stably below 22 degrees is not optimal either from the point of view of equipment operation or from the point of view of the component's operating life, and the danger of high temperatures is sharply overestimated.
So, an article in The Register comes out with a catchy headline traditional for the Register: " Intel says data centers much too cold - Frozen assets a waste of cash" - "Intel claims that data centers are too cold - frozen assets spend money aimlessly."
Nevertheless, the topic is not raised at all "yellow."
The industry-wide reassessment of the danger of high temperatures for data center equipment is confirmed, for example, bya recent Intel study , in which 896 identical blade servers, were divided equally into “control” and “test” groups (8 cabinets, 4 blade chassis in each, 14 blade servers in each chassis, a total of 448 in each site ) The “control” was cooled in the traditional way with the help of a closed-circuit air conditioner, and the test was cooled by the usual “outboard” air, according to the “open” scheme, with its minimum dust removal and without humidity control.
The aim of the study was to prove the possibility of building cost-effective data centers and reduce cooling costs (it is no secret that the costs of a modern data center are very significant. The cost of powering air conditioners can be from a quarter to half in the total energy consumption)
Source:http://dashboard.imamuseum.org/node/830
Vertical - kilowatts, horizontal - days of December. Winter is in the yard, and cooling costs are apparently minimal. The graph does not apply to the Intel test and is given only as an illustration of the typical ratio in the power consumption of a data center.
Also:
Source: http://www1.eere.energy.gov/femp/program/dc_energy_consumption.html
In the case of an Intel experiment, despite the temperature fluctuations in such a "non-conditioned" data center, sometimes rising to 32 degrees, the failure rate compared to the control group differed a little more than half a percent (4.46% versus 3.83% for the "traditional" Intel data centers on average, and 2.45% for the control group, which, in general, is within the "statistical spread").
An even more interesting situation is with the temperature dependence of failures for hard drives. For example, in 2007, a report was published by Google engineers who were studying the frequency and causes of hard drive failures in the server centers of their company (data from about 100 thousand drives were processed, and the study lasted about nine months).
One of the interesting results of that study indirectly confirms the recommendation of ASHRAE, regarding the temperature regime in data centers. So, according to the observations of Google researchers, the probability of failure of hard drives increased sharply when their temperature dropped below 30 degrees, and the lowest probability of failure for the observed group of drives corresponded to a temperature of as much as 40C!
At 40 degrees of operating temperature (all measurements were carried out using SMART sensors), the probability of failure did not exceed 1% AFR (Annual Failure Rate, annual number of failures), an increase to 50C doubled the AFR to 2% (higher temperatures were not observed in the data center) .
But lowering the temperature to 20C paradoxically increased the probability of failures by almost ten times, up to 10% AFR!
On the graph, the histogram bars show the relative number of disks having a particular temperature, points with t-shaped strokes — AFR values for a given temperature and its statistical spread, increasing with decreasing number of “participant” disks with a given temperature.
A noticeable increase in the number of failures at elevated (significantly!) Temperature was observed only for disks with an age of more than three years.
Conclusions:
It is entirely possible that the “the colder the better” approach has outlived itself. This paradoxical, at first glance, conclusion is confirmed by some statistical results, which suggest that, perhaps, we underestimate the “elasticity” of the temperature regime of modern equipment, and its ability to tolerate “elevated” (in our opinion) working temperatures.
In addition, every degree by which the temperature in the data center can be raised there is a direct savings in electricity bills.
First published on the blog http://proitclub.ru/ .