RAID helps if disk is the bottleneck. If it's the network (EBS is network attached), then you're SOL and should probably try a bigger instance instead and hope fewer people share that physical host.
Doesn't Amazon already RAID the storage they offer you? Is drive failure something I need to worry about? I thought they had a beefy RAID system already in place.
They do, but I think he's talking about doing a raid0 just to leech more io, not safety.
I see now with all the comments that everyone is referring to getting more platters spinning the data.
If you enable striping (RAID0) for performance, how do you manage snapshots and backups? Seems like there would be consistency issues re-building the array from multiple snapshots.