Ask HN: Feedback on OSS Project Idea
I want to learn c++, ultra low latency, and low level Linux internals through an OSS project that provides great value to its users. Aside from learning, the main purpose would be to provide a sense of accomplishment to me while doing good.
I'm thinking of a utility that allows greatly reducing the network latency of any program on linux. It would do this by hooking the socket syscalls of the target program into its own implementations that utilize userspace networking. So a kernel bypass without code changes.
The users would benefit from reduced latency and increased throughput, all without code changes. This would be beneficial in several situations, but for instance algotraders could use it.
I'm aware that F-Stack can do this to some extent, but it's limited and quite hard to deploy. I think I would fork it for this.
What do you think about the idea? Am I missing an existing solution that already does everything needed? Am I missing the mark on the user's needs? without much context, it seems reasonable and you should go for it! then write up what you accomplished and learned for us as you go along, HN loves this kind of content