2 ms·
Yea, it sounds like you guys are pretty much at Amazon's mercy right now, which sucks. I think a better way to look at these types of things is "what could we d
by gfodor 17y ago
Yea, it sounds like you guys are pretty much at Amazon's mercy right now, which sucks. I think a better way to look at these types of things is "what could we do to prevent this from happening again" and make a post about EBS gotchas.
Since it sounds like you guys are talking about straight up file storage I'd guess a good option would be to set up an HDFS cluster and be smart about locality/replication to minimize the latency.
Edit: Oh, and one more thing is take a backup that doesn't involve EBS snapshots. Maybe biweekly dump the entire sucker to S3 or something or have it getting pushed there all the time. This is something we've been meaning to do since the snapshotting capabilities of EBS are still a bit too magical for me to sleep well at night. (To be fair though, they've worked great when we've needed them to.)