Most cloud and hosting companies (ourselves included) are keenly focused on system and datacentre availability and how this translates into a Service Level Agreement (or “SLA”) to customers.
We throw SLA terminology around frequently, “99% uptime”, “5 9’s availbility” and “24x7x365 operations” are expressions we hear all too often.
But what do they actually mean in real terms?
The following list outlines what the levels of availability actually mean in terms of acceptable downtime in any given year of operations:
99% = 87 hours, 36 minutes
99.9% = 8 hours, 46 minutes
99.99% = 52 minutes, 34 seconds
99.999% = 5 minutes, 15 seconds
Offering a “five 9’s” SLA to our clients means that we need to be able to offer a service that has a maximum of five and one quarter minutes of downtime a year which we currently cannot offer given the technology at our disposal and the reliance Windows Server 2008 has on hot fixes and service packs that require a reboot of both virtual guests and physical hosts.
It will be interesting to see how Vnext of Windows Server addresses the update/patch issue if at all.