Hi everybody! I have an expensive CPU action that I want to serve through a public internet endpoint. It doesn’t matter what it does, just that it usually takes some time (seconds) and that it costs CPU (say a few percent of CPU usage). For a real use-case you can think of an expensive hashing algo…

Have you thought about measures to prevent clients from misbehaving? For example with public signin pages, rate limits or tools like ReCaptcha can limit the impact an individual client can have.

I have thought about. Rate-limiting is applied but a DDoS would penetrate anyway. reCaptcha is also applied but, well, Google went down these days and brought reCaptcha with it… which made me think about this. In anycase, I wanted to know if there are other papers on the subject or strategies to co…

How much time can a user wait for his/her response? I mean if it takes 10 seconds normally for a process to finish, can the user wait like 10 minutes for it? Is it crucial for the system to do this calculation in a synchronized manner for the system to do other jobs? I mean, if a user can wait a lo…

This is similar to the process pool approach I believe. There the queue is the process mailbox and here is a persistent queue or an external system queue. In any case it is hard to tune it properly in different environments. I wonder if there is some algorithm to auto-tune it according to available…

There is a subtle difference. The process mailbox is external, so it is more fault tolerant and more analysable. I don’t know if there is an algorithm, but have a look at Horizon package that Laravel provides for its queue system. It has some kind of configuration that may help you or give you idea…

If the job takes seconds to complete in the fast path, then you can consider to use a job queue system like: https://github.com/sorentwo/oban Let the general public to use a normal queue and paying customers to use a higher priority queue. This can scale up to many nodes.

Using an external queue system (be it RabbitMQ, Oban, Kafka or whatever) makes it a lot harder to have synchronous responses to reply to the end user. I could pass the PID of the request as an arg and send a reply later and so on or persist the result somewhere and check it later on, but I think thi…

The work involves crypto and that delegates to NIFs that are built-in the runtime (if I understand it properly).

[image] victorolinasc: Using an external queue system (be it RabbitMQ, Oban, Kafka or whatever) makes it a lot harder to have synchronous responses to reply to the end user. You can easily do this with LiveView.

Elixir Programming Language Forum

Constant CPU cost algorithm

Questions & Help Questions

dom December 29, 2020, 3:52am 9

Is the work done in pure Erlang/Elixir code, a NIF, a dirty NIF, an external process?