RethinkDB: how many connections?

I start with rethinkdb

in python

and look at different approaches:

  • Blocking approach with threads
  • A non-blocking, callback-based approach with Tornado

  • Greenlet based approach using gevent

In the first case, it is natural to provide each thread with an object connection

. In the second and third cases, however, I don't quite understand.

With tornado and gentent, how and when should I create connections? How much should I have?

+3


source to share


1 answer


If you are using a non-blocking library, one connection should be sufficient in RethinkDB 2.0 (up to 2.0 was less for each parallelism connection). However, the connection overhead is quite low. Some people open a connection for every request, and even that is not too slow, so you should just do whatever you want.



EDIT: This advice is outdated. For newer versions of RethinkDB that use one connection per request, it is strongly discouraged. One connection per stream is still fine.

+5


source







All Articles