Centralization, and in some cases outsourcing to off-site hosted facilities, is a current trend that continues to unfold in response to the 2008/2009 recession.
This centralization has placed a tremendous focus on the construction of new, more reliable, and more redundant infrastructure than ever before.
Did you catch the operative words there? "New" and "More" ... Yes, as this centralization brings a greater amount of technology closer together physically, so too does it centralize the consumption of power ... and naturally the generation of heat.
It's easy to take data center HVAC systems for granted. They operate with little ongoing maintenance and accommodate modest ongoing growth without upgrade or change.
Many existing data center HVAC systems, however, may not incorporate the level of scalability demanded by the times. This is especially true for managers of small and medium-sized business server rooms and private, corporate data centers where the HVAC systems were designed around the narrow scope of a traditional client/server environment.
All of this makes now a good time to proactively look at data center HVAC systems to determine the level of upward mobility available.
Christian Perry of Processor Magazine offers an article with tips like using temperature monitors strategically positioned throughout the data center to monitor temperatures over time and implementing power management systems that reduce the stress placed upon HVAC systems.
Remember, infrastructure projects like these should take a well-planned and methodical approach. Allow a few months to track and develop metrics before doing anything. This also gives your team ample time to carefully analyze the alternatives, reach out to specialists, and determine the degree of future expandability to build in to any upgrades.
In cases where improvements are being made to systems already in place, Eric Silva of E&S Construction Engineers in Maryland recommends taking an integrated approach.
Simply implementing a patch to fix a heating problem or otherwise making an improvement without proper planning, he says, "will cause the system to compensate in some other area."
Silva also reminds that engineers who engage regularly in developing HVAC systems in data centers have a broad base of proven expertise they use to routinely develop the most cost effective solution possible.
Silva writes:
Stand-alone solutions are tools that a mechanical engineer would use to create an integrated solution to the problem of a hot spot.
By using these tools, an engineer could rebalance a duct system to eliminate the hot spot, determine whether a hot spot can be allowed in a certain area, or pick a location for a new piece of cooling equipment that could be integrated into a data center's chilled water system.
It is also possible that the engineer could determine that there are simpler devices that could be used, such as cold-aisle containment, which could be implemented at a lower cost.
Let's take a look at adding a new duct to bring cooling to a hot spot. We'll make the assumption that cool air flows through a ducted system (not an open raised floor) for clarity. Duct systems are designed to deliver cool air from the AC unit to the space to satisfy the cooling load. Once the duct system is designed, the fan and motor are selected to move the required amount of cool air through the system. Therefore, ductwork is an integrated system with a limited amount of cool air flowing through it.
If one were to add a new duct to the system, the existing system would become unbalanced. Namely, the existing system would deliver some of its cool air to the new duct, would not supply enough cool air to other areas served by the same duct system, where temperatures would increase.
While it is possible to force more air through the duct system to accommodate additional loads, it would require an increase in fan power, therefore increasing the Power Usage Effectiveness (PUE), and would not provide a directly proportional amount of additional cooling. Even so, the drive for energy and cooling efficiency within the data center has caused modern design to limit the factor of safety (spare capacity) of fans and motors that are selected for AC units and there may not be sufficient power available.
Steve Kolbe is the CEO and founder of ANALYSYS, and Eric Silva is a design engineer at E & S Construction Engineers Inc.
Image CC licensed by Flickr user Claus Rebler.