"The most scalable MQTT Message Broker.
Powering IoT, M2M, Mobile, and Web Applications."
Warning, abbreviation overload! I don't know if it's because of Game of Thrones (which I don't even watch) but I find "IoT" intensely difficult to remember as an abbreviation. And I had to look up "M2M". I guess MQTT is acceptable because people interested in this would already know it.
Interesting open alternative to IBM's MessageSight products. In general, the MQTT standard seems great for providing open, high-volume publish-subscribe APIs (over TCP for native apps, over Websocket for web apps).
I compare this to Faye and the Bayeux protocol which I've been using this far, but which few developers know or are willing to learn.
I wrote a very simple queue for embedded devices to talk to (in Go, so you just copy the single binary), which is similar to VerneMQ (I guess) but way less complex:
VerneMQ committer here, we haven't published any benchmarks yet. I haven't looked at the emqtt codebase in depth, but as far as I can tell they use a different clustering/distribution model than VerneMQ. I also don't know about their take on plugins.
Thanks for sharing. Looked over the code, looks solid, and very well organized.
Perhaps would like to see some benchmarks and comparisons with other brokers (including RabbitMQ plugin).
In general with MQTT I was always wondering about its QoS level 2 -- "Once and one once delivery". How well does that work with distributed systems. That would seem a bit hard to implement. (I can see at least once, at most once. Exactly once would be tricky. Thinking about network partitions, disconnects and other such distributed systems shenanigans).
13 comments
[ 3.2 ms ] story [ 35.2 ms ] threadhttp://webcache.googleusercontent.com/search?q=cache:https:/...
Warning, abbreviation overload! I don't know if it's because of Game of Thrones (which I don't even watch) but I find "IoT" intensely difficult to remember as an abbreviation. And I had to look up "M2M". I guess MQTT is acceptable because people interested in this would already know it.
Same here, though my confusion stems from "IoC" -- Inversion of Control. When I read "IoT" I thik "Inversion of -- Trust? no doesn't fit"
I compare this to Faye and the Bayeux protocol which I've been using this far, but which few developers know or are willing to learn.
http://www.slideshare.net/paolopat/mqtt-iot-protocols-compar...
http://www.iotprimer.com/2013/11/iot-protocol-wars-mqtt-vs-c...
http://blogs.vmware.com/vfabric/2013/02/choosing-your-messag...
I don't know if this is what you meant with "public API".
[1] http://zatar.com
http://www.stavros.io/posts/messaging-for-your-things/
It's basically an HTTP API for polling pub/sub, along with a streaming endpoint for push.
I quick read of both VerneMQ and your queue service indicates that your guess is wrong:
- The whole MQTT spec (compact on-the-air messages, hierarchical topics, wildcard subscriptions, QoS level, message persistence)
- Service level agreement
- Live code upgrade
- Monitoring and tracing built-in
- Extensible with plug-ins
- Fault tolerance
Curious because I was about to start playing around with emqtt
Any benchmarks for VerneMQ?
Perhaps would like to see some benchmarks and comparisons with other brokers (including RabbitMQ plugin).
In general with MQTT I was always wondering about its QoS level 2 -- "Once and one once delivery". How well does that work with distributed systems. That would seem a bit hard to implement. (I can see at least once, at most once. Exactly once would be tricky. Thinking about network partitions, disconnects and other such distributed systems shenanigans).