Safety in numbers

Occasionally this can reach four nines or 99·99% reliability, which translates to about 59 minutes down time per year. However, in the age of the Internet, many organisations need over six nines uptime. This means 99·9999% reliability or just 32 seconds down time per year. This is certainly a tall order.

To achieve six nines or better, one has to focus on two aspects of an operation. First, a good electrical system, designed with adequate redundancy and back-up, is paramount. This can be achieved by installing an uninterruptible power supply (ups) system, adding redundant units, dual-feeds, static switches or back-up generators. Secondly, once a data centre is built it must be run and maintained to ensure that reliability targets are met and the operational goals are achieved. If the appropriate steps are taken at the design stage, running and maintaining the centre will be much easier. How well the centre is maintained is ultimately the deciding factor in actually achieving uptime goals.

There are three approaches to maintenance, each with a dramatically different impact on system reliability. The first, and least advisable, of these is to effectively do nothing, allowing the system to crash before repairing it. This is what occurs in organisations with no genuine maintenance policy and is an all too frequent occurrence.

Alternatively, one might opt for a preventative maintenance system, designed to minimise crashes. Preventative maintenance programmes are based on a timetable or schedule not unlike changing the oil in a car every six months.

Finally, there is the predictive maintenance programme. This is a continuous improvement programme accomplished by using metrics to answer certain, vital questions, such as is my power good or bad? Is it getting worse? Where is it getting worse? Why is it getting worse? Ensuring that power quality improves is the ultimate goal in taking a proactive approach to avoiding disruptions and down time.

Monitors are installed to provide post-mortem data when electrical systems fail. This is done in order to understand what happens in any given event and what can be done to avoid a repeat occurrence. This is a minimum requirement; monitors should always be installed to provide post-mortem data. This is a vital part of predictive maintenance. It could be described as functioning in a similar way to an aircraft's black box. Indeed, the Federal Aviation Administration of the US Government's requirement for black boxes in aircraft is part of its philosophy of proactive continuous improvement. It is much more than a mere device to record past errors.

Rhopoint Systems and its partner company, Reliable Power Meters (RPM), offer monitoring instruments that provide post-mortem data. In addition, these systems provide early warning of incipient problems so that action can be taken to avoid them.

RPM has developed a technology for comprehensive power monitoring and performing predictive analysis, which is capable of continuous, real time, predictive analysis on electrical systems. Implementing sound operational and maintenance policies that embrace the philosophy of continuous improvement is the only path to success.

An approach that will yield more nines of uptime must include monitoring and predictive maintenance. RPM achieves this with advanced monitoring instruments and predictive analysis software. The RPM Power Recorder can capture any event, of any size and length, automatically. This is a major advance over older monitors that are complicated to program or require trigger levels and thresholds to be set. Since it is impossible to predict the size, length or number of events in all situations, older monitors often miss important data. RPM has solved this problem with its Full Disclosure technology. Where other monitors are blind to subtle deteriorations in power conditions, RPM's Full Disclosure technology monitors every heartbeat, and every murmur on the power line. Anomalies that are initially benign but grow and become serious in time, can be tracked and identified before they become disruptive. RPM's predictive analysis software, Scenario, analyses the data from RPM monitors, then uses algorithms to trend and compare data over time, thus highlighting any problems. RPM's Power Quality Index (PQI) is the metric that analyses events and provides a simple mechanism to determine if the power is deteriorating. The PQI will identify if and where power is getting worse. By comparing the PQI to other parameters, Scenario will identify why power is getting worse. Armed with this information, action can be taken to avoid disruptions.

Scenario can communicate to multiple monitors in a single facility or in remote locations by Ethernet and via TCP/IP. Sites anywhere in the world can be monitored via Intranets or the Internet. Modem communication is also available. Data can be displayed against worldwide or local maps, installation photographs of facilities, or line diagrams. Easy-to-use zooming tools isolate individual monitors and their power measurements.

RPM's polling software automates data collection and provides alarms to alert key personnel to problems using various methods such as pager, fax or e-mail. Alarms can also be passed to building management and security systems, or to other environmental monitoring systems, to provide a common platform for notifying key personnel.

In many operations, downtime can cost over half a million pounds a minute. Surely it is therefore not prudent to operate a mission-critical facility without a power quality monitoring system that can perform predictive analysis.