Uptime Funk You Up? Don't Get Caught Out By Complacency

So Uptime, or how long something provides service for, is important - and the longer the better, right?

This is just a few musings on why we look at long uptimes and have worries when we see it.

First, we all know about 99.999% (or 5 nines) availability which is the amount of unscheduled downtime should be less than 5 minutes per year. This is a good thing, but one factor that people regularly miss  is that ‘unscheduled’ word. 

So 525,600 minutes is a year, and if we see an uptime counter of greater than 525,595 this would look good to the untrained eye. It means it’s been up nearly a year, minus those 5 minutes. However, when we look at this three things spring to mind.

• When was this device last patched?
• Are there any bugs in that code base that increase in chance after long uptimes?
• Has anybody actually looked at this device recently?

Having an availability SLA of 99.999% does not mean you can’t reboot the unit ever, just that you have to schedule it. And if you’re not patching regularly, you aren’t meeting the minimum maintenance of the device.

Patches clear unexpected issues, and the associated reboots help. Most of them only get found when software is in a wide deployment under both extended runtime and high stress workloads. 

It comes down to restarting on your terms, or eventually, it’ll restart on its own terms.

As always, if you need advice or assistance, we're always here for you.