| Re: [mosquitto-dev] A decentralize Mosquitto cluster design. |
|
sure, I'll do that later. From: mosquitto-dev-bounces@xxxxxxxxxxx <mosquitto-dev-bounces@xxxxxxxxxxx> on behalf of Tatsuzo Osawa <tatsuzo.osawa@xxxxxxxxx>
Sent: Tuesday, January 16, 2018 20:55 To: General development discussions for the mosquitto project Subject: Re: [mosquitto-dev] A decentralize Mosquitto cluster design. Hi Jianhui, Oegma,
Thank you for your reporting.
Could you test the scenarios bellow?
1) Compare with single node, 2-nodes, 3-nodes, ...
- I am interested the overhead of scalability, because it will be the fact of the effectiveness of this architecture. I
don't need to compare with bridge, because I think this is not the alternative of bridge.
AND
2) The best case and the worst case.
2-1) Each subscriber connect with the same node connected by publisher.
QoS=0, No retain.
2-2) Each subscriber connect with the other nodes connected by publisher.
QoS=2, Always retain.
2-3) The moderate case in need.
AND
3) Measure the limit of total messages/sec.
- With my experience, the CPU usage of mosquitto doesn't always represent the performance.
Regards,
Tatsuzo
|