Every time AWS, Heroku, GitHub, Twitter is down it gets posted here and this is exactly the discussion the takes place:
Commenter1: Well they should've never relied on X platform. Single point of failure, etc.
Replyers: It's not Platform X's fault, the engineers should've set up servers like XYZ. It's not like it's hard to do.
Then someone says their app is down, someone else says this sucks, some statistics on uptime get thrown about, and the exact same conversation ensues as the last time any other service went down.
So my honest and sincere question is this: why? Why do "Service X is down" posts get posted and upvoted to the front page? Why do we all have the same discussion each time it happens? It just seems like there's just nothing else to say about it and these sites have status pages where we can all find out for ourselves if they're down or not, so why are we repeating ourselves?
Heroku, AWS, GAE are popular and well-known... but by tightly integrating an application or service to a proprietary platform, you're pretty much at the mercy of the PAAS provider.
To avoid vendor lock-in, check out OpenShift[1] from Red Hat. It's open source[2] so you can set it up on your own servers if you felt Red Hat weren't providing the service you required, or as redundant servers for fail-over.
I thought Heroku was committed to building out more robust handling after the last AWS issue? Rather disappointed with Heroku's uptime solely based on articles I've seen on HN.
I'm not a customer of Heroku, I've stuck with Rackspace the last 5 years because they are located in San Antonio and I'm in Austin. However, many project I've worked on the last 12 months use Heroku for production.
I would think that as such a platform, building compatibility for multiple providers would be a top priority. Beyond compensating for technical difficulties, it would let you shop around for the best price and performance at any given moment.
They've already built their own complex scaling system that doesn't seem particularly tied to AWS. With a moderate amount of additional work, I don't see why they couldn't deploy it to any provider with an API.
Amazon is showing "Performance issues" on their status page.
I think we've all misunderstood. They actually aren't reporting service disruptions, but rather their revenue having having "performance issues" with all the SLA crediting they'll have to do.
Our production site has been down for an hour (a separate staging site on heroku is still up and running), and it does appear to be Elastic Load Balancer-related from what I can tell.
Hugely pissed. My SSL site has been down since 10pm EST last night. It's a huge traffic day for us as well. Damn those AWS and Heroku jackasses. No excuse for this nonsense occurring almost regularly. I'm done with Heroku for good. You'd think AWS and Heroku would figure out some kind of way to handle it when AWS-East goes down. But I guess Heroku is too busy overcharging people to care. /rant
How can they handle anything if AWS-East goes down? Heroku is only using AWS-East.
Don't get me wrong, I wish they would use other AWS regions as well. As a European I really want them to utilize AWS Ireland. But currently there is no need to bitch about how they are handling AWS East outages as there is nothing to handle.
Well, I am very sorry to hear your stuff is down, but I'd think you'd have learned this lesson a while ago. From what I have heard about Heroku they set out to build an easy to use app platform (which they did). Something like that is hugely complex to deploy. My suggestion would be if you must have a PaaS, service, you start looking at open-source alternatives like Cloud Foundry or Openshift.
[+] [-] bpatrianakos|13 years ago|reply
Commenter1: Well they should've never relied on X platform. Single point of failure, etc.
Replyers: It's not Platform X's fault, the engineers should've set up servers like XYZ. It's not like it's hard to do.
Then someone says their app is down, someone else says this sucks, some statistics on uptime get thrown about, and the exact same conversation ensues as the last time any other service went down.
So my honest and sincere question is this: why? Why do "Service X is down" posts get posted and upvoted to the front page? Why do we all have the same discussion each time it happens? It just seems like there's just nothing else to say about it and these sites have status pages where we can all find out for ourselves if they're down or not, so why are we repeating ourselves?
[+] [-] bitcartel|13 years ago|reply
To avoid vendor lock-in, check out OpenShift[1] from Red Hat. It's open source[2] so you can set it up on your own servers if you felt Red Hat weren't providing the service you required, or as redundant servers for fail-over.
[1] https://openshift.redhat.com/community/paas
[2] https://openshift.redhat.com/community/open-source/download-...
[+] [-] kh_hk|13 years ago|reply
Everything worked pretty nice!
[1] https://sscgol-h2o.rhcloud.com/
[+] [-] druiid|13 years ago|reply
And then the 'pay' version created by Activestate and supports PHP and a pretty GUI http://www.activestate.com/stackato
[+] [-] vampirechicken|13 years ago|reply
[+] [-] neya|13 years ago|reply
[+] [-] pxlpshr|13 years ago|reply
I'm not a customer of Heroku, I've stuck with Rackspace the last 5 years because they are located in San Antonio and I'm in Austin. However, many project I've worked on the last 12 months use Heroku for production.
[+] [-] TillE|13 years ago|reply
They've already built their own complex scaling system that doesn't seem particularly tied to AWS. With a moderate amount of additional work, I don't see why they couldn't deploy it to any provider with an API.
[+] [-] redegg|13 years ago|reply
I think we've all misunderstood. They actually aren't reporting service disruptions, but rather their revenue having having "performance issues" with all the SLA crediting they'll have to do.
[+] [-] rtdp|13 years ago|reply
Not just because of AWS, but on there own they have many issues with platform.. a short list - https://twitter.com/rtdp
we need to seriously shift away of heroku asap !
[+] [-] manaslutech|13 years ago|reply
[+] [-] ovechtrick|13 years ago|reply
[+] [-] recurser|13 years ago|reply
[+] [-] rohit6223|13 years ago|reply
But status emailing system is really good :)
[+] [-] manaslutech|13 years ago|reply
[+] [-] briandear|13 years ago|reply
[+] [-] sleepyhead|13 years ago|reply
Don't get me wrong, I wish they would use other AWS regions as well. As a European I really want them to utilize AWS Ireland. But currently there is no need to bitch about how they are handling AWS East outages as there is nothing to handle.
[+] [-] druiid|13 years ago|reply
[+] [-] wilfra|13 years ago|reply
Our game is on Heroku and it's up http://warsocial.com