"Tim" == Tim Keller via WLUG <wlug@lists.wlug.org> writes:
Tim> What do people of the crazy atlassian outage? They're talking Tim> about some customers still being offline for two weeks! Live by the cloud, die by the cloud... I think they seriously screwed up, which goes to show you the programming is 10% the work to be done, and 900% error, bounds and exception handling. All of which computers do poorly, but which we humans do reasonably well. But we're *slow* and the cloud is all about speed. In this case, it also shows that Atlassian's DR practices were crap in a major way. And if I was self hosting it internally, and knew that I was being forced to the cloud... I'd be yelling like hell at my sales rep and working to get off Atlassian products. The cloud has some advantages, and is lovely if you have a work load that scales horizontally really well, and which also bursts up in terms of performance need. But when you have a fairly static demand, it's maybe not the best thing to do. But ... the cloud sells itself on doing all those hard things well, like redundancy, resiliency, durability, etc. dual feed UPS, tested generators, reliable AC systems, etc. All the things a good Co-Lo or self hosted people handle to some level of execution. The cloud should (ideally) have all this down to a T. But ... when it doesn't, all hell breaks loose. John