Last month the us-east-2 RDS region had a major outage which took our multi-AZ primary database cluster down for 4-5 hours. We spend the entire outage online with a support agent but basically had to just wait on their team restoring full service as there was no way to stand up a new cluster or transfer a snapshot out of the region due to the outage.<p>Even though the RDS SLA makes this severe outage a very small 10% credit we went through the motion and followed their instructions for requesting an SLA credit with documentation of our affected cluster. We're now at 15 days since our request and they've come back twice in that time, the first requesting the information already included in the original request, and the second requesting that we attach a spreadsheet calculating what we think our SLA payout should be.<p>This process has left a pretty bad taste in our mouth regarding AWS service and their treatments of outages and SLAs. 5 years ago I would have said AWS was pretty good at this stuff and recommended them. I'm just wondering if anyone else has recent anecdotes either way. Is AWS slipping or becoming scummy in general or is this an isolated incident?
I think it is a by-product of their culture.<p>We recently had an SLA request to them that was for a paltry amount, nowhere near the damage they caused, less than 0.1% of our monthly spend. It was really just to gauge their process and how supportive they would be. Despite our account manager saying they would sort it, it was denied. They agreed we were impacted, they agreed they caused it, they agreed it met the duration, but the service team used a technicality in the wording of the SLA to get out of responsibility. I won't say much more to keep things somewhat anonymous.<p>Our account reps were on our side and are helping us escalate a complaint through on this.<p>We have several ex AWS folks in my team, and they said they weren't surprised. No one at AWS admits to mistakes as it costs promotions and puts you potentially on the chopping block. That's why the request was denied, that's why support is hard to deal with, that's why the status page is never updated quickly. They do a lot of things well and can be decent to deal with provided you understand they won't own up to their faults going into things
Don't suppose you can share a link to details of the outage? As is typical for AWS I find, it's very difficult to find any reference to past outages beyond the default 7days PHD provides.<p>In answer to your question though, I have generally found AWS pretty good when it comes to issuing SLA credits.