Hi,
we're using AppEngine to hosts our multiplayer games. They've got a lot of users around the world and I think we're spending a considerable amount of money on hosting them.
Our games (and I think most of the applications) use different kind of server requests destinations such as FrondEnd workers, static and BlobStore content download. For instance we've used 3M requests in a couple or hours today.
The financial credit form ask for the percentage of failing requests from a total of requests. I think that's not fair cause we only need a BlobStore service down to have our complete games out.
To clarify this, suppose an application sends 100 different requests when starting. If only one of them fails, the complete app will fail. But the AppEngine way to measure this is only 1% because other 99% requests were working good.
I know that some requests could fail and you must retry it. But by downtime I mean that they will keep failing for more than a minute or two.
Am I wrong on this? What do you guys think?