I have to preface my answer with a quick disclaimer. My most relevant experience was with RabbitMQ and NATS, whichHow is this different from other cloud based messaging frameworks, such as MQTT for example?
is not exactly MQTT and the mentioned experiences were not in a WAN context. However I did some quick reading and
there are comparisons made between RabbitMQ and MQTT.
The MQs (RabbitMQ, NATS and MQTT) at their core adopt a store-and-forward approach to delivery of a message from
a subscriber to zero or more publishers. The intermediate storage is at the foundation of associated guarantees of delivery.
Their goals are to achieve highly reliable delivery of messages to potentially large numbers of receivers (MQTT makes
claims of connection to millions).
This is fundamentally different to ansar. An intermediary is used to stitch two transports together such that a block
sent into the local end of leg 1 pops out at the remote end of leg 2. The block are never stored. There are no claims
about connection to millions of devices. It might be possible to connect to tens of thousands. The primary goal of
ansar was to make network messaging easy without losing sophistication. Implementing a "universal send" happened
to expand into the WAN space.
This does not mean that ansar cannot be used to deliver highly reliable services. It has adopted an SDL-style approach
to network messaging which was the basis for many highly reliable protocols in the telephony space. This forum is
probably not the right place to go much deeper than that.
Thanks for the question. I may add something to the docs.
Statistics: Posted by mr-ansar — Tue May 28, 2024 9:43 pm
