I don't like S3 for reasons of redundancy and cost. At the high levels mentioned above S3 becomes quite costly and if you're efficient you can be looking at break even points in favor of owned infrastructure in less than a year.
This is followed by the cost of trying to meet your own SLAs and then compounding that with 3rd party services like S3 who fail and bring down your services with them.
Cost and SLAs I understand, but redundancy? I would think part of the reason to put it in the cloud is so you don't have to have the data mirrored across multiple machines.
And even for SLAs I'm not sure it's that bad. Any homegrown solution is going to have similar stability problems. It's on machines running in data centers, and there are only so many things you can control.
It's really important, however irrational, to be able to tell your boss you're "working on it" when something breaks. That means you'll favor the reliability you can control over the reliability you can't even when it may be lower!
You can buy almost 3x the storage hardware with 1 year's worth of storage cost, to say nothing of the transfer cost. Then it all depends on the economies of scale. A small company's incremental operations cost approaches (or in some cases bests) Amazon's as the number of servers under management grows. Once you're in the 10s of TB, a good ops team can soundly beat S3 in ROI. Opportunity cost is a different story, (maybe those precious engineering resources were better spent elsewhere) though opp cost is highly situation specific.
Comments 4
Reply
This is followed by the cost of trying to meet your own SLAs and then compounding that with 3rd party services like S3 who fail and bring down your services with them.
Reply
And even for SLAs I'm not sure it's that bad. Any homegrown solution is going to have similar stability problems. It's on machines running in data centers, and there are only so many things you can control.
Reply
You can buy almost 3x the storage hardware with 1 year's worth of storage cost, to say nothing of the transfer cost. Then it all depends on the economies of scale. A small company's incremental operations cost approaches (or in some cases bests) Amazon's as the number of servers under management grows. Once you're in the 10s of TB, a good ops team can soundly beat S3 in ROI. Opportunity cost is a different story, (maybe those precious engineering resources were better spent elsewhere) though opp cost is highly situation specific.
Reply
Leave a comment