Availability is all about what they will pay for

Back when I still got involved in LinkedIn discussions I got caught up in a discussion re SLA levels (or strictly speaking SLT levels). What I noticed was I didn't see much mention of the BIG question: what is the customer prepared to pay for?

It's all very well to survey what they need and what the business impact of an outage is etc etc But five-nines costs many times what four-nines does. Likewise stepping from three-nines to four.

Or from 85% to 95% :D

The foundation of an SLA is a two-way negotiation.
"We want this."
"Great. It'll cost you that."
"Oh. Well how about we have this instead then."

Secondly, what can you deliver? If development changes the app every week, if the operations staff all smoke dope, if the management have all been there for thirty years and wear cardigans, if the business insists on users having admin rights... then you aren't gonna hit five-nines at any cost.

Third, what doers the customer need? Never mind what they want. "Ask a customer what they want and they'll say they want ponies". What is the genuine business requirement? What problem or need or risk is that addressing?

We aren't writing theoretical frameworks here. This is the real world. A bit less theoretical navel-gazing and a bit more getting back to business realities would be a good thing in many of these ITIL discussions.

P.S. Five-nines doesn't mean 5 minutes outage a year, not unless you are providing 24x7x365. It means 99.999% of the agreed availability window. That window should be subject to at least as intense discussion as the subsequently agreed availability level. "24 hours a day? Really? Not 23.5? Or 20? Well 24 will cost you this much more..."

Syndicate content