How Adaptive OTA Updates Work Without Extra Infrastructure

Key Takeaways

Could you talk a little bit about the approach uh to delta updates that Rugx takes? Yeah, so there are different ways. There's this like static delta compression where you take an artifact and you take the new version and then you statically compute the delta uh between them. Um and then there is kind of these adaptive delta updates which is what rug is doing. So the implementation works as follows. You have your update. You chunk this update into blocks. Then you put a block index into the update which basically says okay these are the blocks that make up the file system. You do the same thing locally on the device also compute these blocks and then at runtime uh the client installing the update will look at uh like which blocks are locally available and which don't I have. So I have to fetch them and then it will using HTTP range queries fetch those parts adaptively that it does not already have locally. And generally with static data compression there are cases where you get like a higher compression ratio. So you get smaller updates but then of course you have this this whole infrastructure around that that you need. And the nice thing about the the adaptive data updates really is that you can just put the artifact somewhere like in an S3 bucket or it doesn't really matter as long as it supports HTTP range queries and it just it just works without so much infrastructure.