Monitoring in the data center seems to bring out a wide range of opinions from industry figures. Those responsible for capital projects tend to see monitoring as a big target when value engineering costs out of a project. After all, they reason, points and devices can always be added later and besides who could possibly look at all the thousands of points. On the other hand, those responsible for operating sites tend to want to monitor everything with every change of state immediately emailed to their Blackberry.
In the last several years, equipment manufacturers have converted their standards to utilize electronic controls, and our abilities to monitor just about everything right down to CRAC units and 15-amp branch circuits have improved dramatically.
Does it make sense to monitor over 40 operating parameters on a simple chilled-water CRAC unit?
Likewise, on a branch circuit it may make sense to monitor the actual amperage and create alarm states at 75, 80, and 90 percent of rated load but does one really need to do waveform capture at this level? Yes, someone actually asked for that! We tend to be pack rats when it comes to collecting data about the data center operation but seldom utilize that data to its full benefit.The following is one man’s opinion as to what is reasonable and necessary.
Utility Level. Every utility serving the site should be monitored for quality and trends. The most sophisticated utility is electricity; incoming switch gear includes waveform and disturbance analyzers. Other utility services such as water service and quality are just as critical (high pH, solids, bio-organisms, etc) but are often forgotten
Distribution and Equipment Level.At this level, it is important to get into what type of client you are and what you really need to know. A health-care provider with lives at stake will be in greater need than a research computing facility running batch processes. Monitoring is further defined by the data center size, infrastructure design, and staffing levels.
As health-care providers go totally digital, they increasingly depend on their data facilities. The computerization of health-care provider services has already produced significant improvements in patient care, reductions in prescription errors, enhanced response times for test results, and a lower cost of service with fewer repeated tests.
As this digital health-care world expands and becomes more reliant on digital content, monitoring of the data center becomes more critical and therefore justifies a greater investment in monitoring key operating parameters of the power and cooling distribution.
Although research operations tend to have heavy capital investment, they have a propensity to have an established equipment environment (few changes) with extreme power densities. Once they are up and operating these environments tend to be quite stable and predictable. As a result, their monitoring needs tend to be quite basic.
Enterprise data centers tend to fall between health-care and research operations and to extreme examples with transactional and process industries having more monitoring. So what are the basics and what are the enhanced monitoring parameters for distribution? Here is just a sampling:
The major difference between basic and enhanced monitoring is that basic monitoring is limited to real-time whereas enhanced monitoring provides operating data that can be trended and analyzed. As data centers become more densely populated with less and less tolerance for out-of-spec environmental factors, we will need to develop failure predictability software. Many advanced sites already are doing this by setting various alarm thresholds and user notifications when key parameters are approached or exceeded. At the top end some will even reset emergency operating parameters when back-up equipment is out of service.
Enhanced monitoring is opening new doors. Now we can trend and predict load growth based upon real data. We can analyze loads by application, by department by processor, by disk type, etc. If data are properly collected, we can use the operating data of installed 1U servers against the equivalent capacity of blade servers to analyze the power and cooling impact of converting from one technology to another and other various what if scenarios.
So now back to the original question. Monitoring: Is it worth it? The answer is an obvious yes, but the value we receive has as much to do with how we use the data collected as the data itself.Email comments email@example.com
Next Issue: We are running out of power. Then what?
About “Cronin's Workshop”Win a Mission Critical coffee mug
The enormous creativity of our industry means that we face radical changes in the way we design, build, operate, and maintain mission critical facilities. Making this change will demand strong debate about what is right, wrong, and best practice.We have created “Cronin’s Workshop” to foster this debate. Each month our columnist Dennis Cronin will describe a problem and pose a solution. We want to hear what you have to say about these problem/solutions posed by Dennis. In the subsequent issue we will publish the best positive and negative opinions received from readers. Please email contributions to the editor firstname.lastname@example.org