The data center should neither a sauna nor the Sahara be. Environmental sensors can help monitor optimal temperature and humidity levels to keep the data center equipment working reliably.
Safety controls such as fire suppression equipment and water detection sensors also need to be located properly and routinely tested to protect the company’s investments. Our expert Bill Kleyman shares his expertise on the proper data center environmental monitoring techniques in this Q&A.
Q. What parameters should be part of a data center environmental monitoring scheme?
Bill Kleyman: A data center environment monitoring scheme can take many different forms. It all depends on the type of organization. Many companies demand high levels of security and visibility into the server room’s physical environment. An administrator should try to monitor common environmental factors including:
- Temperature: No server or server rack should operate above a maximum inlet temperature. Many administrators look for rack air flow exhaust metrics, internal system temperatures and even CPU temperatures. The more information that is provided by the temperature control mechanisms, the faster an engineer can nip issues before they become serious problems.
- Humidity and water control: Water leaking from
- chiller lines or other sources can damage electronic systems. Similarly, excessively humid air can condense into liquid water that poses an equal danger to servers and systems. Deploy water sensors and humidity sensors at strategic locations both inside and outside of the rack.
- Aisle conditions: Containment has emerged as an important part of cooling management, so administrators should distribute environmental sensors within the hot and/or cold aisles. This can offer a broader picture of what’s happening within a cooling zone.
- Fire suppression monitoring: Many large data centers have a comprehensive fire suppression system, but these systems can sit idle for years. Will they work if a fire starts? Short of testing the mechanisms, administrators should have an active monitoring system to ensure the equipment is working.
- Static electricity sensors: Excessively dry air, as well as improperly grounded equipment or personnel, can accumulate potentially harmful levels of static charge. Static electricity monitoring equipment located at intervals around the data center can report on the presence of potentially large charges.
- Server room and rack entry: This is all about physical security: doors opening and closing. Room and rack entry sensors send alerts if the data center or rack has been improperly entered. Advanced environments can activate a camera that points to the rack where the cage has been opened.
Q. Are there any established guidelines that specify the server room environment? Certainly ASHRAE makes recommendations for temperature and particulates, but are there any industry standards for overall environmental conditions?
Bill Kleyman: It’s hard to get specific with environmental monitoring best practices because each server room has different sizes and equipment and, therefore, different requirements. However, there are core environmental conditions that should be observed. The factors you measure will depend on the size and complexity of the infrastructure.
- Temperature: Gauging temperature will always be a key component. The optimum range for equipment stability had been recommended between 70 to 74 degrees Fahrenheit (21 to 23 degrees Celsius). In 2008, ASHRAE increased the range to 59 to 89.6 degrees Fahrenheit. In 2011, ASHRAE raised the limits again — for the very latest equipment — to 41 to 113 degrees Fahrenheit.
- Humidity: A humidity sensor is fairly standard in any sized environment. Relative Humidity (RH) is the ratio of moisture in a given sample of air at a given temperature when compared to the maximum amount of moisture the sample could contain at that temperature. The recommended RH is between 45% and 60%. More humidity can condense into liquid water, while less humidity can cause electrostatic discharge (ESD) — either scenario can damage equipment.
- Water: There should be no liquid water standing in a data center, so sensors at low points within racks and aisles can report on leaks and condensation problems.
- Airflow: Maintaining good airflow is vital for both temperature and humidity control. Good airflow recommendations vary depending on the size of the environment, and are expressed in volume as cubic feet per minute (cfm). The actual volume of air required is related to the moisture content of the air and the temperature difference between the supply air and return air. Avoid turbulent airflow as this will be felt as a draft. This is where size of the environment becomes very important: In a dense server room, the number of air changes per hour may be several times greater than that of a smaller environment.
- Computer room air conditioner/handler: CRAC units must be monitored constantly. This includes supply and return temperatures, internal humidity statistics and air-loss percentages. Any faults in this unit need to be dealt with immediately.
- PDU and electrical system status: Electrical circuits within the environment should be watched continuously for unexpected fluctuations or disruptions. Any disruptions would be a serious emergency (not to mention a potential hazard to the facility). In addition, the data collected from power distribution locations is often used to calculate power usage effectiveness (PUE).
Q. What about environmental sensors? How durable and reliable are today’s environmental sensors? Should sensors be maintained, tested or replaced, and if so, how often?
Bill Kleyman: Environmental sensors really report on the “health” of the overall facility, but no sensor is guaranteed to work forever. That’s why it’s important to have a redundant sensor environment. Intelligent data center monitoring tools will observe all of the sensors and can look at multiple sensors at the same time in case one has failed. Sensor redundancy can eliminate false-positives when a sensor fails. Environmental monitoring also requires careful alerting: if a sensor fails, the right administrator and technician must be notified promptly. This is the same if the device starts to post incorrect information or triggers false alarms.
Q. How should environmental sensors be placed? Are there any tools to help optimize sensor placement in and around servers, or is this still a manual hit-and-miss process?
Bill Kleyman: Since each environment is unique, there are no tools to determine the optimum sensor placement. A certain amount of empirical trial-and-error is a normal part of environmental sensor placement. However, working with a data center and HVAC professional can help an organization plan the best deployment. There are a few general placement guidelines:
- Temperatures will be greater at the rear and top of a rack, so measure exhaust temperatures and air flows there.
- Temperatures will be lower at the front and bottom of a rack, so measure intake temperatures and air flows there.
- Humidity will be highest in warmer air which can hold more moisture, so place humidity sensors higher in open areas away from any distinct heating or cooling sources.
- Place water sensors in low locations such as along rack bottoms or near drain openings, usually below any raised floor.
Q. Let’s talk about integration. How do environmental monitoring tools interoperate with administrator-focused systems (server) management tools and facilities-oriented building management tools? How do these all fit together to give companies a complete picture of what’s happening in the data center?
Bill Kleyman: Large data centers must have clear visibility into the entire environment. This isn’t just environmental information — this means server metrics as well. There are tools to watch power consumption, CPU, RAM and other vital components in conjunction with environmental monitoring systems. For example, AVTECH Software Inc. provides a variety of appliance-based monitoring tools as well as sensors for comprehensive monitoring. Other tools such as up.time Software will help an administrator monitor distributed data centers and gauge resource utilization.
Ultimately, how well a large infrastructure runs is due to the communication between the data center teams, not because of the tools. Alerts must go to the correct engineer and manager from server, data center and virtualization teams, in a coordinated effort to create an optimally functioning environment. Data center consolidation has been an ongoing initiative with many organizations. This means that larger servers are supporting more workloads. Teams with monitoring capabilities must collaborate with one another to create an environmental diagnostic plan should an event occur with any system. Integration of major systems should be done with a provider capable of handling the environment and organizational needs of the customer.
ABOUT THE AUTHOR: Bill Kleyman, MBA, MISM, is an avid technologist with experience in network infrastructure management. His engineering work includes large virtualization deployments as well as business network design and implementation. Currently, he is the Virtualization Architect at MTM Technologies Inc. He previously worked as Director of Technology at World Wide Fittings Inc.
This was first published in February 2012