DCIM systems for temperature monitoring alone makes little RoI sense for server rooms.
It’s hard to believe but many companies with smaller in-house data centers, those with 10-20 racks, and with server rooms containing a handful of racks and telecom equipment are often not automatically monitored for temperature. Although numbers are difficult to quantify, the fact is that many suppliers like Temperature@lert have a significant IT piece of their business. So it’s not hard to understand that DCIM suppliers whose primary pitch to large data centers is asset management and whose RoI in smaller data centers and server rooms is not asset management but environmental monitoring. After all, dedicated temperature monitoring devices do not require the built in capability or expensive software packages that DCIM product necessitate.
Why is this so? Taking a look at a recently published application note a major DCIM company discusses how to follow ASHRAE data center temperature operating guidelines using RCI (Rack Cooling Index), RTI (Return Temperature Index) and multiple RFID (Radio Frequency ID) sensors coupled to a RF reader plus asset management software as an interface to the data adding up to several hundreds if not thousands of dollars of investment and manpower to install, train and use the system. Contrast that with a dedicated WiFi temperature monitoring device capable of supporting up to four wired sensors and which has all of the software needed pre installed set up to send email and capable of supporting text messaging and SMTP authentication, that provides text and XML log temperature readings for easy integration to custom applications that costs a few hundred dollars. The RoI on the dedicated device will be a fraction of the time as the DCIM system. But can it do the same job?
ASHRAE data center recommended temperature range, 2004 and 2011 update. Link to Source. ASHRAE recommended temperature are is between 68–77 °F (20–25 °C), with an allowable range spanning 59–90 °F (15–32 °C).
Large data centers generally have full environmental and security monitoring systems to meet their uptime and security goals. Companies running Cloud providers know within seconds when something goes very wrong and soon after the entire web connected universe is alerted when their favorite hosted applications are not available. In 2013 Dropbox, Facebook, Google Drive and Facebook suffered self inflicted outages while Amazon’s cloud service took down websites Reddit, Github and Minecraft among others. This year Dropbox, Samsung, Adobe Creative Cloud, Microsoft, Verizon and Time Warner Cable suffered major outages and it’s only August when this is being written.
Not all of these were due to temperature related failures. According to 2013 data in the table below 27% of data center failures reported in the survey were “Heat related CRAC failures”. This is a significant number given that data centers designed for optimum cooling and does not bode well for small server rooms where AC is often tied into the general building HVAC where in hot summer months the demand for personnel comfort can help unbalance the air distribution and raise server room temperatures significantly. Couple that with the all too frequent AC unit failures and the need to monitor server rooms seems like a no-brainer.
A comparison of data center failure causes in 2010 and 2013 shows Heat related CRAC failures noted by 27% of respondents. UPS battery failures led the list with two categories noted and human error came in as #2. (Link to Source)
The DCIM piece notes that raising the temperature of a “typical 8,000 sq.ft. data center can save at least $64,000 per year in cooling costs by raising the temperature by 2°F while remaining safely within the 2011 ASHRAE temperature envelope guideline window. To implement this rack and return air temperature cooling were measured in several locations by placing multiple RFID temperature sensors with four RFID sensors for every third rack plus a
Rack Cooling and Return Air Temperature are two categories to monitor according to one RFID monitoring system supplier. That’s a lot of equipment and cost. Add to that installation, training, personnel to use the data to then perform experiments to achieve the energy reduction without jeopardizing uptime or availability, and maintenance of the battery powered sensors.
For server rooms the good news is that room temperatures have been operating relatively safely within a safe operating range. At most sites things have been working smoothly for some time and it’s only when the AC is interrupted that problems occur. So what is needed to monitor such an event? The answer is a simple WiFi device capable of supporting up to four wired sensors for a few hundred dollars. The sensors will be located near the front of the HVAC outlets. The two other ports can be used to monitor temperature near the ceiling where heat will accumulate most quickly. Users will need to take data near the HVAC inlet and outlet to understand the range of temperatures in order to set the alarm levels, and to monitor the ceiling level sensors for a period of a few hours to do the same. The trick is to find the balance of normal high and low temperatures before setting the upper and lower limits. This will help avoid possible nuisance alerts when temperatures drift near the upper operating range, on hot summer days for example. An offset of 5 °F (3 °C) from the upper and lower operating range is a good place to start. The range can be tightened up or loosened over time if needed. And energy saving experiments can be done with a close watch on the readings to achieve the savings promoted by the vastly more expensive system.
Temperature@lert WiFi Edition server room monitor (left) vs RFID environmental monitoring system described by RFID supplier (right). https://www-304.ibm.com/software/brandcatalog/ismlibrary/details?catalog.label=1TW10MA2V
The choice for small data centers and server rooms is clear, thousands of dollars and time for a system whose primary application is asset management vs a dedicated, self-contained WiFi temperature monitoring device for a couple of hundred dollars. Not only will the WiFi cost be significantly lower but installation and setup will be accomplished in a fraction of the time for RFID. And managers will like the easy justification and RoI.