During the great Amazon Web Services outages of April 2011 and June 2012 many users were stuck. They could, in theory, move their apps to another AWS region, or to another cloud provider altogether. But in practice the architecture of AWS regions are unique – each one supports slightly different features
and APIs. Update: It’s not as quite difficult to move between regions as I thought, see comments for more details.
The thing is, every cloud provider will eventually have some downtime. Quite possibly less downtime than your on-premise apps. But if you can’t wait it out, you’re going to need to architect your applications so that they can live in different environments. And today AppFog and Rackspace announced a partnership that promises to make it easier to do that. But there’s no word on when, so for now this is still vaporware.
AppFog is a platform cloud, like Heroku or Google App Engine, that supports several programming languages and frameworks. It’s based on the private platform-as-as-service software Cloud Foundry, which VMware open sourced last year. So far AppFog has only been available on AWS’ infrastructure, but CEO Lucas Carlson has been promising a one click migrate between clouds for a while now. What AppFog announced today is that its service would be available through the Rackspace Cloud Tools Marketplace. Its inclusion will bring the the platform-as-a-service marketplace I described last weekend a bit closer to reality.
Seamless app migrations between clouds is a tall order and I’ll believe it when I see it. But it would be a killer app for platform clouds, which so far have left many developers asking “What’s the point? I can configure all this stuff myself.” But AppFog’s not alone, the platform market is increasingly crowded. For example, just this week Uhuru, another Cloud Foundry based provider that adds-on .NET support to the core offering, launched its beta. Cross-cloud deployments are likely on the roadmaps of many, if not most, platform cloud companies.
If nothing else, hopefully this move will put some pressure on AWS to make it easier to fail-over to other regions.