Hello Hirchel,
Thanks for the link. It provides a nice view of why you might want to switch to ACE on containers with a full orchestration environment too, but doesn't really address what happens if you run in a bare container on a VM.
So, what does the Integration Node do for you, and what would you have to build to replace that functionality if you wanted to run containers on a bare VM?
Off hand, the Integration Node does the following, and probably does a bunch more things that are less obvious, or that I haven't run across lately.
The Node orchestrates startup of multiple Integration Servers.
The Node can restart failed Integration Servers.
The Node can work with MQ to provide Multi-Instance support, allowing automated recovery to a different system in the event of a server failure.
The Node can host the HTTP listeners, and load balance HTTP requests across available Integration Servers.
In a container based world, the container orchestration system (commonly kubernetes) does all of this sort of thing. It manages startup and multiple copies, load balancing, restart on failure, and adds features like dynamic scaling based on load.
However, if you wanted to run stand alone containers directly in a docker engine, without container orchestration, you would need to build these things yourself, or live with the manual processes and potential outages related to stop/start, failure recognition and restart, server failover, and if you wanted to have multiple Integration Servers providing the same services with load balancing, you would need to implement the proxy server and load balancer to enable that capability.
------------------------------
Neil Casey
Senior Consultant
Syntegrity Solutions
Southbank
+61 (414) 615-334
------------------------------
Original Message:
Sent: Mon April 08, 2019 04:59 PM
From: hirschel wasserman
Subject: Integration node followup
Just noticed this link https://www-01.ibm.com/common/ssi/cgi-bin/ssialias?htmlfid=15017915USEN
------------------------------
hirschel wasserman
Original Message:
Sent: Mon April 08, 2019 04:54 PM
From: hirschel wasserman
Subject: Integration node followup
HI
There was a previous post around the purpose of using a Node in a non-docker VM world.
From Ben: "For users who are not using containers, you can run integration servers in VMs or directly in the software installed "on the metal" of a machine. In these circumstances, the servers should be affiliated with an integration node which "looks after them" ... ie The node process starts the server process and restarts them should it be needed. The node can also be configured to be highly available across physical LPARs etc. too. This approach (using servers which are owned by a node) is the traditional approach ... "
Could I ask that someone unpack the highlighted statement into greater detail? As we move toward migrating to ACE, we are trying to reach decision around instantiating an Integration Node or not. (VM non- Docker).
If we chose not to standup a Node, what additional processes would we have to put in place.
Thanks
------------------------------
hirschel wasserman
------------------------------