Hi Dave, I don't think there's a particular limit to it. For a single application, the idea is to share a single Pulsar client instance. That makes sure to reuse a lot of resources (threads, TCP connections, etc..)
For each producer, the biggest amount of memory would be related to the queue of outstanding messages. The default for the queue size is 1000, but it could be reduced in case the required throughput for each producer is is not that high. I haven't done the real check on the amount of memory (I don't expect it to be super-huge), though it could be tested using the `pulsar-perf` producer from CLI. The tool can easily simulate N producers on N topics. Matteo On Wed, Nov 1, 2017 at 1:28 PM Dave Fisher <dave2w...@comcast.net> wrote: > Hi - > > I am doing some design work for a message based workflow service I am > thinking about building. > > What is the limit on the number of simultaneously open topic producers on > a client node? > > Regards, > Dave > -- Matteo Merli <mme...@apache.org>
