When I asked about "What does it take to satisfy you when testing if a service is available or not?" the topic of SLAs came up. Specifically, user MSimmons said "Re: What does it take to satisfy you when testing if a service is available or not?
That got me thinking about SLAs - specifically the times that I've been let down by them. My latest example is a VPS that I keep around for my business. The VPS has had poor uptime because of the service provider's networking devices that are in front of it. One time it went down in a bad way. Like, almost two days of being unavailable. Suddenly I couldn't rely on it for client-facing operations. My business's reputation was tarnished. Clients were then wary of using that part of my offering. Things got a bit uncomfortable. "No problem!" I thought. "I've got a SLA!"
As it turns out, my SLA has a tiered structure of credited service. 10% downtime results in a certain percentage off of my next bill. 20% increases the percentage, and on until, at a sufficiently high percentage of downtime, my account gets a completely free billing period.
Awesome! So now I don't have to pay for a service that I can't rely on! That's... just... super. What's worse is that if, as a result of the downtime, I perform a hasty migration from the service, I'll get a credit on my account to go towards the next billing period... except I won't be using their services anymore so I'll essentially get nothing for my trouble.
How do you negotiate SLAs? What are some gotchas that you've encountered? Is there any specific provisos that you've learned to either be wary of or demand be included? How do you ensure that you'll get adequately compensated for the downtime that a service encounters? How do you handle remuneration from an SLA? Do you have any good (or bad) SLA stories?