When Services Die


Summary

What happens when your favorite service dies? Will your data live on? Not if you're dependent on cloud-based services you don't control.

flickr Badges

Last week I was speaking to Jeff Kramer about services that go away and how personal clouds could help solve that problem. (Incidentally, Jeff has a nice post on archives that he put up earlier this week.) Jeff mentioned a talk from Aaron Cope called Time Pixels. The talk is museum-centric since that's what Aaron does, but it's message isn't just about museums.

Aaron hits on a few ideas that resonated with me. The idea of giving things unique identities is important. He takes it to an extreme of giving things unique, decodable identifiers based on Hilbert spaces so that the identifier could be specific to not only a particular lat-long coordinate, but the time too. More importantly, Aaron talks about what can we do about services that go away. He makes a proposal, that solves the problem from a society (museum) standpoint. I have one that is aimed at individuals.

I started thinking about this when I started using Instagram. I realized that Flickr is likely to die sometime. I don't want to debate whether Flickr will die because every online service will die. It's only a matter of time. As Jeff said to me, we're really good at creating new things and not very good at keeping them going. I know people like Doc Searls who put a lot of pictures on Flickr. What's he going to do when Flickr dies? Presumably he has his pictures somewhere else too, but a large swath of what Doc does with his pictures is tied up in Flickr.

Naturally, being an engineer, I started to think up solutions. And because I'm steeped in the ethos of personal clouds and personal data, my solutions ran in that direction. Assume for a minute you have a personal cloud. And also assume that you have configured your personal cloud to use online storage that you control. The cloud provides an abstraction for managing that online storage so that it's portable. You can chose Dropbox or S3 or Rackspace or your cousin Eddy's hosting service. You can even decide to self host. You can change your storage provide without losing functionality. In fact you might use multiple storage provides simultaneously for backup or reliability.

Someone could write an application for your personal cloud that uses a service like Flickr or Instagram for taking, sharing, and indexing pictures. But all the file storage would be in your personal cloud rather than on their servers. This is easy to imagine and not even hard to implement if you have a personal cloud. If the service goes away, you not only still have the pictures, but you also still have the app and the functionality it provides. By moving functionality, especially storage, out of services that are silos and into a service that you own and control, we preserve important data regardless of whether the service provider lives or dies. We also have an easier way to deal with privacy—but that's a different blog post.

Personal clouds represent a significant change to how services are delivered to users. The personal cloud architectural has significant advantages leading to better personal control of data resulting in privacy by design.