For datacenter operators, plenty of change is in the air. They have the ever-shifting and generally increasing demands for processing power and storage capacity. There are the swelling prices and shrinking supplies of power. Moreover, there's forthcoming change to environmental legislation that could have a profound impact on how -- and where -- a company bases its datacenter operations.
I had an opportunity to hear firsthand how one datacenter operator is dealing with these changes during a visit to RagingWire, a managed datacenter host in Sacramento that happens to be SMUD's (Sacramento Municipal Utility District) largest customer. During the tour of the 12.6MW, 110,000-square-foot datacenter, James Kennedy, RagingWire's facilities manager, shared with me some of the strategies he's undertaken to cut energy consumption and costs associated with running the datacenter. Moreover, he discussed future challenges the datacenter industry is facing and how current and forthcoming environmental regulation will alter the datacenter landscape and drive cloud computing.
Like any cost-conscious IT company, RagingWire cares about uptime, performance, and keeping energy costs down. To that end, the company has invested in an array of monitoring tools to track the performance of its systems, as well as the environmental conditions of the datacenter. Its tools of choice include Liebert SiteScan, which monitors and controls Liebert precision cooling, power, UPS, leak detection, control panels, and other equipment; GE's iFIX, a SCADA (Supervisory Control And Data Acquisition) engine; and SynapSense, which employs wireless sensors to monitor environmental conditions in real time.
These tools prove helpful in tracking of the datacenter, equipping RagingWire with real-time alerts when problems arise; current data to calculate such metrics of PUE (Power Utilization Effectiveness) on a regular basis; and the ability to generate reports to present customers with a breakdown of the performance and costs of their respective leased sections of the datacenter.
[ PUE is a useful metric -- but the Uptime Institute warns not to put too much stock in it. ]
These tasks aren't entirely automated; you still have to figure out PUE yourself, for example. As monitoring tools continue to evolve and communicate with one another, expect to be able to figure out the PUE of your datacenter -- or even just one rack in your datacenter -- in real time.
These types of tools help with planning for changes in the datacenter to ensure that few watts and square feet go to waste. If, for example, a customer wants to add more servers or storage capacity, monitoring tools give RagingWire an assessment of what sort of power, cooling, and floor space is available, one that is far more accurate than the old-fashioned method of manually gathering data and crunching numbers via spreadsheet for a rough estimate. From there, RagingWire can advise a client to, for example, upgrade to higher-power servers in their existing footprint instead of leasing more space.
Measure it and you can manage it
Kennedy's assessment of his experience with SynapSense's wireless environmental monitoring system was of particular interest to me. (I wrote about the company and its product offering last year.) SynapSense is a relatively new company -- it's been around for a couple of years -- but has already landed customers circling the globe. Moreover, HP, which once touted a datacenter monitoring system of its own called Dynamic Smart Cooling, now offers the SynapSense system as an OEM.
One of the most obvious benefits of SynapSense is its ability to monitor environmental conditions in real time, providing meaningful, useful data on temperature, humidity, and other conditions. A datacenter operator can use the information to view a facility map and determine where hot spots or other environmental anomalies can be found. From there, the operator can take action -- and that's where the payoff occurs.
For example, Kennedy explained that RagingWire seeks to maintain the ASHRAE-recommended temperature for the datacenter, the highest temperature possible that still ensures systems won't overheat. RagingWire then needs to adjust the set temperature of the facility to compensate, for instance, when chillers are being restarted (during which the datacenter will grow warmer for a brief period of time), as well as the mixing of hot and cold air.
By mapping and monitoring the environmental conditions of the datacenter, RagingWire has been able to identify problem areas and fix them. Adjustments have included installing blanking tiles to improve airflow and balance pressure, hanging curtains at the ends of server aisles to reduce mixing, and taking other steps. As a result, the company has managed to safely increase the CRAC temperature of the facility from 55 degrees to 71 degrees, and adjust the temperature of the chilled water -- used for cooling -- from 44 degrees to 60 degrees. The blanking panels alone have reaped the company a savings of $400,000, according to Kennedy. Increasing the temperature just from 55 to 60 resulted in savings of another $200,000.
Kennedy noted that another selling point of SynapSense is its flexible wireless design. A datacenter -- particular a hosted datacenter facility where equipment is constantly added and moved -- is a dynamic place. Hence, datacenter monitoring tools need to be easy to move and easy to scale. Picking up a wireless sensor and attaching it to a different rack is far easier than uprooting a hardwired piece of hardware and reinstalling it elsewhere.
Carbon regs mean datacenter shifts
These types of tools are essential for monitoring and maintaining the efficiency of a datacenter, helping keep costs low. But it also helps in reducing RagingWire's environmental footprint, a measurement with which datacenter operators in the United States and abroad are becoming increasingly concerned as environmental regulations aimed at reducing GHG emissions continue to evolve and emerge.
For example, the United Kingdom has already embraced a carbon cap-and-trade system; lawmakers in the United States are considering doing so as well. Meanwhile, the Environmental Protection Agency is working on an Energy Star efficiency standard for datacenters. Moreover, Kennedy told me that RagingWire is in the process of being permitted as a utility by the EPA and California Energy Commission, as it has backup power generators at its location.
The shift toward stricter environmental regulations will have a significant impact on the datacenter industry, Kennedy says. For starters, it will mean higher operating costs for datacenter operators as they'll be essentially paying a carbon tax on their energy consumption. That, in turn, will be passed on to customers.
Second, Kennedy says that carbon regulation will reverse the trend of datacenter operators leaving California in search of lower energy costs. The reason: A datacenter operator may pay less per watt on the East Coast than in the Golden State, but utilities out there tend to generate their electricity with coal, which is far dirtier than the natural gas used by California utilities. As such, datacenter operators to the east will have to pay higher carbon-related fees -- high enough perhaps to force them closer to a source of cleaner energy. The alternative would be moving their IT operations to a country that doesn't have stringent environmental regulations.
A third and related trend: The onset of environmental regulations, coupled with the cost and complexity of building and maintaining a dynamic datacenter environment that can suit an organizations' current and future needs, will induce companies to abandon the datacenter business entirely and instead embrace the service of a hosted provider such as RagingWire or Digital Realty Trust -- as well as cloud computing services such as Amazon Web Services and Elastic Compute Cloud.
[ Learn why cloud computing is greener than hosting your own datacenter. ]
This, in turn, is going to further drive the evolution of cloud computing, Kennedy predicts. As more organizations rely on managed IT services based miles away instead of in the same building, they'll need their applications to be securely available in the cloud.
So where is this all heading? Kennedy jokingly (or perhaps half jokingly) said that he would like to have the ability to monitor the datacenter from the golf course. That scenario likely isn't too far off. Datacenter monitoring and management tools are growing increasingly sophisticated. If you have access to those tools via the cloud, you should be able to receive alerts -- be it on the golf course or in your headquarters in Portland, Ore. -- on a portable device that temperature in your Santa Fe, NM, datacenter is rising. You could then call up a real-time map of the datacenter and see that a CRAC unit is down. From there, you can notify your staff in Santa Fe of the problem and perhaps turn on a backup CRAC unit with the click of your mouse, all without leaving your office -- or your caddy.