Differences

This shows you the differences between two versions of the page.

Link to this comparison view

irc:1468188000 [2017/05/27 13:44] (current)
Line 1: Line 1:
 +[20:00:32] <​ChicagoJohn>​ for a microservices architecture,​ would you ever use the service proxy or the event bus?  or is it primarily for HTTP rest calls?
 +
 +[20:02:15] <​ChicagoJohn>​ the microservices examples (aggregate and pipeline) only make use of HTTP calls. ​ follow up, is service discovery something you use alongside microservices? ​ i saw you have the docker discovery.
 +
 +[20:05:48] <​dbh613>​ event bus is great for microservices receiving some event/​message for which they need to process, doing something, and possibly putting another event on the message bus for further processing
 +
 +[20:07:59] <​dbh613>​ In the cases I've used it, I haven'​t used a service proxy or HTTP calls. ​ In low-memory / slow CPU / single core instances where the microservices need access to results from others, ​  ​I'​ve used  mutliple verticles (microservices) running in the same vertx instance, and used Shared Memory
 +
 +[20:08:51] <​dbh613>​ Event Bus  (with HazelCast behind it)  was too memory overhead ​ for  a Raspberry Pi B+ with 512 MB ram,  overclocked single CPU core to 700 Mhz
 +
 +[20:09:00] <​dbh613>​ too much memory overhead
 +
 +[20:13:28] <​ChicagoJohn>​ lets say we do microservices in docker containers. ​ between those containers, would that imply only HTTP?
 +
 +[20:14:31] <​ChicagoJohn>​ from what i gather from your raspberry work, you deployed multiple independent verticles on to the same machine and used the event bus.
 +
 +[20:22:35] <​dbh613>​ ChicagoJohn:​ Those microservices (in different docker containers) could communicate over the event bus
 +
 +[20:22:52] <​dbh613>​ Unless you have some need for it go over some other protocol
 +
 +[20:23:34] <​dbh613>​ the event bus allows different instances of Vertx  to discover each otehr and share/​communicate with each other
 +
 +[20:23:52] <​ChicagoJohn>​ is that because of the docker service discovery '​plugin'? ​ or is that just inherent in vertx to be able to discover each other because they are ion the same subnet
 +
 +[20:24:39] <​dbh613>​ event_bus
 +
 +[20:24:42] <​dbh613>​ http://​vertx.io/​docs/​vertx-core/​java/#​event_bus
 +
 +[20:25:06] <​ChicagoJohn>​ thanks
 +
 +[20:25:18] <​dbh613>​ Hazelcast discovering other instances of hazelcast on the same subnet
 +
 +[20:25:33] <​ChicagoJohn>​ science man....science
 +
 +[20:25:37] <​ChicagoJohn>​ thanks
 +
 +[20:25:41] <​dbh613>​ There is also support for Apache Ignite as the event bus
 +
 +[20:26:42] <​dbh613>​ all of that messaging can be  pub/​sub ​ or  point to point
 +
 +[20:28:55] <​dbh613>​ It is pretty slick stuff from what I've seen so far.