How Slowcast built an efficient client without the burden of a full fledged backend cluster.
Castro 2.0 is coming soon to the App Store and the hype is building up. Supertop recently released a blog post on how they built Castro 2.0 server cluster to handle the heavy lifting to achieve a more efficient client. In this post, I want to show you how Slowcast achieve pretty much the same client side efficiency without the burden of a “fat” backend.
The subtitle is actually a little bit deceptive since Slowcast is taking a totally different approach as most of the existing podcast clients. Instead of eagerly push/pull updates to/from the client, we pull based of the pace of how user listen to their subscribed podcasts. Subscriptions in Slowcast is basically smart bookmarks which will auto download the next episode when you almost finished the existing one. With this design approach in mind, we can build a more efficient client without a heavy weight backend. Instead of going through every single subscription and check if there’s new updates, we only need to keep track of what you actually listened. For average users, that means 1–2 pulls per day. Doing this allow Slowcast users subscribe to a much wider range of podcasts without the cost of high networks or disk consumption.
We consider podcasts as more of a “on demand” medium rather than a “live” medium, so we think that the way to consume podcasts should be designed to fit this property. With the design principle to minimized information overload, we are able to build a more efficient client with less server side burden. We hope that by building this opinionated podcast client, we scratched some users itches.
You can download Slowcast at App Store. Please let us know what you think. :)
