News for hackers. Stuff that matters.
Re:Heroku's AWS outage post-mortem by RyanKearney on Tuesday April 26, 2011 @10:41PM (#2488164)
What about this example. You have Acme Corp Datacenters who sell dedicated servers to their customers. If Acme Corp has a network outage because their single Comcast connection went down because Comcast was having some routing errors, the customers who are effected go to Acme Corp. It isn't the customers fault that Acme Corp wasn't prepared to deal with a downed connection and setup a redundant network.

In this example, think of Amazon as Comcast and Acme Corp as Heroku. Heroku wasn't prepared to handle this type of failure, so they're at fault.

Re:Heroku's AWS outage post-mortem by tzs on Wednesday April 27, 2011 @01:35AM (#2488519)
By that argument, the customers of Heroku who weren't prepared to handle Heroku's failure were at fault. They should have had alternate rails hosting lined up.
Re:Heroku's AWS outage post-mortem by jasonlotito on Wednesday April 27, 2011 @12:02PM (#2489930)
Amazon is at fault to Heroku. However, Heroku isn't talking to Amazon here, they are talking to their customers. Heroku is at fault to those customers. In turn, those customers are at fault to their users. Heroku can't pass the buck to Amazon, and those customers can't pass the buck to Heroku for their downtime (They can, but it's still their responsibility).
Re:Heroku's AWS outage post-mortem by RyanKearney on Wednesday April 27, 2011 @02:07AM (#2488567)
Well it depends on what Heroku's SLA was, if any. If Heroku stated 100% uptime grantee, then Heroku would be to blame for not living up to their 100% uptime. If Heroku said hey listen, we can't guarantee any amount of uptime so be prepared and someone were to host some "mission critical" information on Heroku then yeah it would be the customers fault.
Re:Heroku's AWS outage post-mortem by railsguy1 on Wednesday April 27, 2011 @03:41AM (#2488758)
Heroku has no SLA