A
A
Andrew2014-04-29 18:24:54
ruby
Andrew, 2014-04-29 18:24:54

Is there an alternative to regular TCP sockets for a multithreaded parser?

Task: collect information about ~100 million pages on different domains as quickly as possible using the HEAD method.
Now the parser is written in ruby, uses HTTPClient and multithreading.
Maybe there are faster libraries for such tasks?

Answer the question

In order to leave comments, you need to log in

2 answer(s)
S
Stanislav, 2014-04-29
@crackedmind

1) Run on jruby/rubinius to get real multi-threading for distributing threads across cores, otherwise the GIL interferes
2) Run multiple processes

S
Sergey, 2014-04-29
Protko @Fesor

Well, let's start with the fact that HTTP uses TCP as a transport (at least from the server side), so ... no.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question