Your datacenter is an expensive way to keep your Pepsi cans cool!
Guest blog from Srikanth Aithal - Senior Staff Software Engineer at IBM India
Once, we visited a customer for an IBM Systems Director enablement engagement. He wanted to take us to his datacenter to show all the servers and other datacenter equipment. We followed him to his datacenter. As he was the only person having access to the datacenter door, he requested access and held the door open for us to get inside. The datacenter was not so big, it was around 2000 sq. ft., and the temperature inside was very cold.
I had a query related to the systems the customer had, so I turned back to ask him the question. There was a surprise shocker for me to see the customer wearing a monkey cap (Balaclava). I asked him the reason for wearing it, he told me that he had bought that monkey cap to keep his head warm in the datacenter. He also cracked a joke telling me that the set point temperature in the datacenter can also keep his Pepsi cans cooler. Definitely, they had oversized the cooling capacity for that small datacenter with the current IT equipment that it hosted.
It was my understanding that they had oversized the cooling capacity considering future growth, but the current set point temperature seemed very low. If you are wondering what is the set point temperature, it is the temperature at which the Computer Room Air Conditioning (CRAC) aims to keep the internal air temperature of a datacenter.
What can make a datacenter temperature rise? The simple answer is mainly the IT equipment hosted in datacenter consumes power for computing and releases heat which increases the datacenter temperature. The CRAC units are in place to remove the heat to keep datacenter temperature at set point temperature. The closer the set point temperatures are aligned with the outside external temperature, the less energy the CRAC needs to operate.
The lack of insight into the actual power consumption of IT equipment hosted in the datacenter can lead datacenter administrators to make assumptions on the set point temperatures needed to eliminate the heat generated and keep their datacenters at normal temperatures. Energy and thermal measurements can be done at different levels to understand the overall thermal map of the datacenter and arrive at a realistic set point temperature. A couple of them are done:
At the overall datacenter level, including facility and IT equipment
At the individual IT equipment level, including servers, storage and switches
IBM helps clients to understand their overall datacenter level thermal maps using tools such as mobile measurement technology. To understand the thermal map of each individual piece of IT equipment, first we need to be aware of their actual power consumption. Many times, we take the nameplate wattage of the IT equipment as their real power consumption values; which can result in the calculation of a higher thermal map; therefore, leading to more cooling capacity requirements.
Active Energy Manager (AEM), which is a plug-in to IBM Systems Director, can help us to understand the actual IT equipment power consumption. AEM, also help to save power consumption on the IBM Power 6 and later processor based-servers as well as cap the power usage on supported IBM servers. The AEM monitoring feature, which can monitor power, thermal and other environmental values on IBM servers, non-IBM IT equipment and legacy IBM servers is free to our clients.
Recently, I was involved in the writing of the IBM Redbooks publication, IBM Systems Director 6.3 Best Practices, SG24-8141-00
. Included in this IBM Redbooks publication is a chapter on Active Energy Manager. Do not miss reading about AEM to get started with Power monitoring and management in your datacenter, which can help you get rid of monkey caps.
Srikanth Aithal Senior Staff Software Engineer at IBM -
Connect with me on LinkedIn
Image provided by Nelson Velazquez - Twitter handle: @nelsonvelazquez