Excellent commentary on the situation. My company is currently pricing cloud hosting for a system-as-a-service product and EC2 is one option we've considered. The recent outage certainly didn't improve anyone's opinion here at the office, but I don't think too many people realized the outage so bemoaned on Twitter was really something that could just as easily have been a minor nuisance. Because of past experiences, I suspect we'll likely go to Rackspace Cloud and pay their premium prices in exchange for not having to deal with the setup intricacies of EC2.
I think that was really the biggest issue. People don't know how to plan for real availability, and when disaster strikes, they're left wondering wtf just happened.
EC2 almost makes it too easy to deploy an application with no forethought.
Can I put in a solid recommendation for Storm on Demand (Icrontic's host)? Not only is LiquidWeb reliable, but their customer service has been stellar. In addition, good ol' Ardichoke is an employee there and we've gotten lots of very personal and awesome service from him and his colleagues.
Here's the real question, though: Are we load-balanced between or among their datacenters?
That's the problem that hit most of the people who got hammered yesterday. They had all of their resources in one or two zones within the region. When those zones had problems, they were dead until resolution.
I've personally recommended Storm on Demand to many people but based on the scale of what we're doing and, most importantly, previous experiences I don't see the decision makers swaying from what they know to work.
Doesn't look to me like it's back up. Amazon Elastic Compute Cloud (N. Virginia) is still reporting as down on the status page. Also, http://www.cad-comic.com/ is still down (the only reason I even noticed that EC2 was having problems, other than reading about it here and whatnot).
I love the fact that the cad-comic page is "hosted" by ZeHosting, which is one of hundreds of "hosting companies" that just resell cloud services. Even better, ZeHosting's tagline is.... well:
I wasn't able to check in on Foursquare yesterday morning for a couple hours. But, I often have errors on 4SQ on my phone, so it took me awhile to realize it was anything other than business as usual.
Are we load-balanced between or among their datacenters?
No, that would be a massively unjustifiable cost given our size.
Here are my hosting experiences summarized:
Random Rackspace Tech: "Ohai guyz, I restarted ur Apaches, all done."
Me: "You just took down half our websites by making a blatant configuration mistake. Nice."
(This month they are 0/6 in getting tickets right the first time! What a great game.)
--
LiquidWeb Tech I Recognize: "I made the requested change, could you confirm? Also, are you the guys we did X & Y absurdly complex things for flawlessly for last summer? I recognized your hostname."
Me: "Yes. Yes we are. "
AlexDeGruven
Wut?Meechigan Icrontian
Comments
EC2 almost makes it too easy to deploy an application with no forethought.
https://www.stormondemand.com/
(PS: The times Lincoln pushed dev code out to the production server DON'T COUNT)
That's the problem that hit most of the people who got hammered yesterday. They had all of their resources in one or two zones within the region. When those zones had problems, they were dead until resolution.
Also, thanks for the love Brian, we <4 you too.
Here are my hosting experiences summarized:
Random Rackspace Tech: "Ohai guyz, I restarted ur Apaches, all done."
Me: "You just took down half our websites by making a blatant configuration mistake. Nice."
(This month they are 0/6 in getting tickets right the first time! What a great game.)
--
LiquidWeb Tech I Recognize: "I made the requested change, could you confirm? Also, are you the guys we did X & Y absurdly complex things for flawlessly for last summer? I recognized your hostname."
Me: "Yes. Yes we are.