Data Center Efficiency: The Benefits of RCI & RTI (Part 1 of 3)

I continue to be surprised how few data center operators know about the Rack Cooling Index (RCI) and Return Temperature Index (RTI) metrics and how it can benefit the efficiency of their data center.  In my opinion, utilizing the RCI & RTI metrics should be a key part of any strategy to optimize data center energy costs.  Understanding RCI and RTI doesn’t take a rocket scientist, but it does take a little time. 

In order to understand RCI and RTI, you first need to understand the problem that RCI and RTI addresses.  The focus of RCI and RTI is to help data center operators understand and balance the competing requirements of equipment reliability and the cost of cooling.  Everyone knows that if the air intake temperature for IT equipment is too hot then IT equipment reliability suffers.  On the flipside, if the air intake temperature is too cool then energy costs for cooling the data center skyrocket.  Gartner has stated many times that for every 1 degree F that air intake temperatures are raised, 2% of the annual power costs can be potentially saved.  Let’s translate this into actual power costs and savings in a few examples:

• A 20,000 square foot data center with an annual power cost of $2.1 million dollars could save the following:

• 1 degree F increase = 2% savings = $43,449.00 annually

• 2 degree F increase = 4% savings = $86,899.00 annually

• 3 degree F increase = 6% savings = $130,348.00 annually

• A 140,000 square foot data center with an annual power cost of $7.6 million dollars cold save the following:

• 1 degree F increase = 2% savings = $152,752.00 annually

• 2 degree F increase = 4% savings = $305,505.00 annually

• 3 degree F increase = 6% savings = $458,257.00 annually

As you can see, a simple 3 degree F air intake temperature increase can result in real savings each and every year.  However, the real question is can you actually raise your rack air intake temperature and not endanger your IT equipment?   The vast majority of data center operators don’t know the answer to this question.  

ASHRAE (American Society of Heating, Refrigeration, and Air-Conditioning Engineers) has created air intake thermal guidelines for data centers.  

Through exhaustive testing and analysis, ASHRAE has determined that rack air intake temperatures need to be between 18 degrees C and 27 degrees C (64.4 degrees F and 80.6 degrees F).  Data centers that are over cooled (below 18 C) are wasting significant amounts of money.   Data centers that are under cooled (above 27 C) are at significant risk for heat related IT equipment failures.   

For optimal savings, the rack air intake temperature needs to be set as high as possible while staying within the ASHRAE “recommended” range.   This sounds easy but in actuality it is not.  The problem is in most data centers cooling needs are not uniform which is why you have hot spots and cool spots.  Fine-grained thermal monitoring is required in order to provide the information needed to optimize the data center air intake temperatures.   The minimal best practice method for thermal monitoring in the data center dictates 3 temperature sensors at the top, middle and bottom of every third rack.      

Once the data center is instrumented with fine-grained temperature sensors per the best practice recommendations, a wealth of raw temperature information will begin to flow.   The problem now is information overload and trying to “process” the raw data into something useful.  In a data center with 200 racks, 3 temperature sensors on every third rack would result in 200 discreet temperature readings. To make this useful, a report would have to be produced to show temperature values that were outside of the ASHRAE “recommended” range.  Once the report is produced, you will have to study the report and produce some type of analysis of the raw data.  This is where RCI comes to the rescue.   

RCI is a measure of compliance with AHRAE or NEBS air intake temperature guidelines and is expressed as a percentage with the maximum value being 100%.  Actually there are two RCI metrics:

• RCI_HI is a measure of the absence of over-temperatures

• 100% means that no temperature is above the maximum recommended

• Less than 100% means the greater the probability (risk) that equipment experiences temperatures above the maximum allowable (hotspots)

•  RCI_LO is a measure of the absence of under-temperatures

• 100% means that no temperature is below minimum recommended

• Less than 100% means the greater the probability (risk) that equipment experiences temperatures below the minimum allowable (over-cooling)

So in the ideal world, an optimized data center would have an RCI_HI of 100% and an RCI_LO of 100%.  The percentage value is simple and easy to understand – like a grade on a test.  The RCI metrics do the analysis of the raw temperature values for you.  

A really nice feature of the RCI metric is that it can be rolled up or “consolidated” as well as present a very fine-grained view.   If a data center is optimally instrumented with sensors (top, middle, bottom of every rack), then an RCI metric can be produced for each and every rack. The RCI metrics can be rolled up to provide RCI metrics for a row and then overall RCI metrics for the data center.  This provides a very powerful view for the data center operators.  

Once the RCI values are available, data center operators can begin the task of optimization.  While it may be as simple as adjusting the CRAC supply temperature values, most likely it is much more involved.   Optimization may mean installing different types of perforated floor tiles, baffling systems, and variable speed fans.  It may require dispersing high-density equipment as well.   However none of these optimization tasks can be accomplished safely if you don’t understand your current thermal intake baseline.   In other words, you can’t optimize what you don’t measure, and you can’t measure what you don’t monitor!  At RF Code, we can help you obtain the RCI metrics with our wire-free sensors and enterprise monitoring software.  Don’t try to optimize your data center in the blind do it the intelligent way by using RCI metrics.    

In next Monday's edition of this blog series we will discuss the RTI (Return Temperature Index) metric in detail.

Rack Cooling Index (RCI) is a Registered Trademark and Return Temperature Index (RTI) is a Trademark of ANCIS Incorporated.  (www.ancis.us).  All rights reserved.  Used under authorization.