What would generally be your expectation for SLA uptimes across the following services in a given month for a 'normal' scenario?
I'm including my expectations... the purpose of this question is to find out what the standard baseline should be in 2009 for setting SLA goals in an IT infrastructure. (Again, in a normal scenario - we're not talking Amazon or Google, but then again we're not talking Billy Bob's Web Site hosted on home DSL either.)
- Power (100%)
- Network / Core Routing / Switching (99.999%)
- Static Files Hosting (99.99%)
- Application Hosting [Single Server / DB] (99.95%)
- E-Mail Hosting (99.95%)
- Complex Application Hosting / Platform [Multiple Servers / DBs / Services] (99.9%)
(For reference... SLA Calculator)
And finally... How much maintenance time is acceptable within a month?
This will depend on a number of things:
Your business requirements? Do you serve only your local area/timezone or are you truly global?
The cost of downtime? What does an hour's downtime cost in terms of non-productive staff, lost web-site sales, reputation?
Your budget to achieve this? As your requirements go up, so does the cost: from on-call & pager allowances and overtime to full time night shift staff, high-availability and/or redundant servers and software and backup sites.
As to acceptable maintenance, are we talking scheduled maintenance, or unscheduled downtime?
Which side of the agreement are you on? If you're providing the service then it's in your interests to make it as easily achievable as possible while still meeting business requirements.