Storage systems come with sophisticated monitoring systems that allow administrators to keep track of everything that goes on inside the box. EMC
alone sells scores of different software packages to manage its own equipment as well as those of other vendors — it has seven offerings to ensure information safety, for example. It also has “phone-home” remote support to catch and fix errors before they result in data loss. Hitachi Data Systems Corporation
, and others provide similar software and services.
But what about actions that occur “outside the box” — in the literal rather than figurative meaning of the phrase? Those multi-million dollar SANs are located inside a physical environment that also needs to be monitored and managed.
Internal Weather Report
Protecting a data center environment means more than just keeping out hackers. Temperature, humidity, fire, water, and intruders can all influence the performance of machines that store and process data. Some storage units report when their own internal temperatures rise, but what about the temperature for the room as a whole?
And while building engineers have systems that monitor the environment and control the air handling and fire suppression systems, these are designed to protect the building, not the data. Joe Balsamello, vice president of the New York City-based data center design firm M. Moser Technology International, relates an example of what occurred at a financial company.
As usual, the company had a sprinkler system in the building that was hooked into the smoke alarms. A belt on an air conditioning fan in the data center ceiling started smoking. This tripped the sprinkler system right over the firm’s main trading server. The machine was flooded before the IT staff could shut it down.
Heat is just as much of a threat as water. The air conditioning system will control the overall room temperature, but what about the air temperature down amongst the server racks where additional cooling is needed?
“We were having problems in one of our server rooms with hot spots, especially with the new close-packed racks,” says Eric Halter, manager of monitoring services for Merrimack, NH.-based Fidelity Management and Research (FMR). “The blade servers were generating hot spots in the room, and we wanted to figure out how to better direct our cooling as well as being able to check for problems.”
The answer to both these problems, and many others, lies in adding environmental probes to the existing IT management system.
Environmental monitors are Ethernet-enabled units that measure and monitor environmental factors and alert administrators whenever a threshold is exceeded. There are several companies that manufacture such devices, including Austin, Tex.-based NetBotz Corporation and Thailand-based AKCP Co. Ltd. The devices come in various configurations depending upon whether you want to put them on a rack, a shelf, the wall, or the ceiling.
Each comes with one or more sensors (for heat, humidity, airflow, voltage, smoke, water, etc.), depending on what the particular site needs. In some cases, you can even put a sensor on the door and hook it up to a camera so that it emails you a picture of each and every person that enters the data center. All of the units come with their own server-based alerting and management software, or administrators can hook them into the company’s network and systems management software, and monitor them as just another network device.
Page 2: Keeping Cool
So, how can environmental monitoring help protect physical storage assets? Taking Balsamello’s example, the smoke detector would not only have set off the sprinkler system, but also could have executed a script to automatically shut down the equipment in the event of a fire.
As another example, you could place one right next to your Symmetrix DMX 3000 to make sure that the air flow is going right where it is needed to cool down those 576 hot disk drives holding 84 terabytes of your mission-critical information. The same device could ensure that the air flow is neither too moist (corrosion) nor too dry (static electricity problems).
In Halter’s case, after considering the probes from both NetBotz and AKCP, he bought a dozen AKCP SensorProbe2 units from top U.S. environmental sensor distributor Javica (Sanford, Maine) — four for each of FMS’s three data centers — since they were simpler to install and more compact. Initially, he just got them to monitor the temperature, but later added humidity monitoring.
He placed the devices in the locations most likely to be hot spots so he could monitor for any problems. He’s also now using them for ensuring better cooling in the design of a new server room.
“Since we can measure the temperature as we move things around, we have been able to correctly place the racks and space them properly,” he explains.
Rather than managing the probes separately, he ties them into his existing Hewlett-Packard OpenView for Operations (OVO) software. He configured it for two temperature thresholds. Crossing the first results in an e-mail warning, while the staff is paged by the software to alert them to take immediate action when the second threshold is exceeded.
“The system lets us know when the air cooling performance starts deteriorating so we can take proactive steps to correct it,” says Halter. “It’s better than waiting for a fan to fail before doing something about it.”
That financial firm with the flooded server could tell you just how much better it is.
Feature courtesy of Enterprise IT Planet.