search

KonduitAI / konduit-serving

Enterprise runtime for machine learning models
https://serving.konduit.ai/
Apache License 2.0
47 stars 15 forks source link

Add HTTPS + HTTP Authentication #230

Open AlexDBlack opened 4 years ago

AlexDBlack commented 4 years ago

This isn't an immediate priority, but it's likely that some users will want some sort of authentication for model serving.

For example, suppose a server is publicly accessible to the internet, serving remote IOT devices (or point-of-sale, etc). Clearly we don't want just anyone being able to access such a server and send inference requests, or query the endpoints (such as to get server metrics).

AlexDBlack commented 4 years ago

Kind of related is the ability to do request rate limiting (for clients) and IP filtering. That might be worth looking at also.

agibsonccc commented 4 years ago

This would likely be kubernetes based. We probably shouldn't implement this till we understand what kind of devices we'll be running on. We don't want to write anything more than we have to. Even then, we'll want to look at what vertx already has available.