In some households, if you only have to hit the TV three times to make it turn on, that means it still works. Service availability used to work the same way, where if a server was pingable, it worked. Those were simpler days when availability dealt with specific areas of infrastructure and machine failure was not extremely uncommon. These days, equipment is built to avoid outages and maintain standards of availability, and downtime has more to do with application or server configuration than network problems. Elizabeth Harrin writes how availability today must take into account every element that contributes toward the user experience.
If availability is to be useful, it must first be defined in a meaningful way so that it can be properly measured. It can be measured as literally the amount of time services are available, or it can also be measured according to response time. If transactions are nearly instantaneous one day and take a few seconds on the next day, the system may be unreliable. And although the old IT mentality of having preset downtime for maintenance is often no longer practical in our globalized work environment, there are now kits that allow for system upgrades without interruption. Not inconveniencing the user is key.
When discussing the cost of unplanned downtime, Harrin gets harsh in her frankness:
Unplanned downtime has a massive business impact. Figures from Alinean show that outages in a messaging system can cost around $1,000 a minute. Downtime for trading applications can cost up to $40,000 a minute, so tolerance of downtime could differ between mission critical and other systems.
It may take some time to define appropriate measures for system availability but there is one thing everyone’s clear on: the opinion of the IT department is not important.
At the end of the day, user satisfaction is the point of availability. Banging the TV might work for you, but if you are going to have the neighbors over, just splurge for the LED set already.