RBCS - Putting Downtime Into Perspective

Putting Downtime Into Perspective

Availability is measured in percentage, and 100% availability of course means being available every second of every day. 99% availability may sound good, but 1% downtime translates to 88 hours in a year, or about 3.5 days. For some security and emergency systems, that is not acceptable. That’s why most systems that bother specifying availability, specify something better than 99%.

Recently my company has reviewed a number of RFPs and specifications that included uptime requirements, but in some cases the planned use of the system didn’t match the high level of uptime specified. Several of the uptimes were over-specified, and several were under-specified based upon an analysis of downtime costs vs. the costs of the fault tolerance specified.

The important task in designing a system's fault tolerance level is determining how many hours the system can afford to be "down" due to unexpected hardware faults, and weighing that cost against investments in fault tolerant hardware and software.

Even for a small to medium security system such an analysis can be worthwhile. For example, if a complete system failure means that 15 security officers must be brought in to manually secure the facility at $300.00 per hour, 24 hours of downtime would cost $7,200. Under that scenario, an extra $3,000 investment in fault tolerant hardware starts to make sense. Of course for some systems, like emergency 911 call systems, the cost of even one hour of down-time can be immeasurable and should be prevented practically at any cost.

Here is what the uptime percentages mean in terms of downtime hours:

	99%	88 hours downtime per year (3.67 days)
	99.5%	43 hours downtime per year
	99.9%	9 hours downtime per year
	99.99%	less than 2 hours of downtime per year
	99.999%	a little over 5 minutes of downtime per year

These are the primary factors that contribute to downtime:

Planned Downtime

Preventive Maintenance

Backup

Repair

Upgrades

Unplanned Downtime

Hardware Failure

Software Failure

User Error

Environment Failure

Generally the higher the uptime requirement is, the higher the cost. If you require less than 9 hours of downtime from the vendor system, yet you plan to take the system offline yourself for 3 hours each month for backups and preventive maintenance, the 9-hour requirement for the vendor may be too strict. If you are dealing with a CCTV monitoring system whose cameras are not on an emergency power system, it’s not practical to specify a 4-hour UPS for the monitoring computer since the cameras will be offline anyway.

When you make the determination of how much uptime you expect or require of the system, make sure that your planning considers all the downtime factors listed above.