Layer7 API Management

Hi expert

One of our customer is planning to run 2 LAC server in parallel without clustering.
Both LAC server has its own each MySQL DB and infront of 2 lac server Loadbalancer is exist.
Both LAC server and DB will have same data.

I did try in my environment. it works fine. but i couldn't understand different between cluster and without cluster.

Is this approach is good?

Any adivice is much appreciated.

Thank you in advance.

Shreeram

With your reference of a MySQL database as the admin database, I am assuming that you are running on version 4.1 or lower. If you are having a load balancer in front of the two LAC nodes, you are running as a cluster. You only need one admin database where one or more nodes of the cluster will all be configured to point to the same MySQL admin database. Otherwise, they will be out-of-sync or you will end up deploying multiple times. Do note that when you import an API to a cluster, you will need to restart all nodes. 

Let me know if there are other follow-up questions. 

Hi chewa03 

Thank you for your quick response.

I am using LAC version 5.1  and MySQL DB version 5.7

According the bellow link, i am assuming that LAC nodes need to configure hazelcast setting for clustering.

please correct me if my assuming is wrong. 

Configure Live API Creator to Run as a Cluster - CA Live API Creator - 5.1 - CA Technologies Documentation 

I am not configuring this options setting, assuming running as a stand alone on both nodes.

In this case, I think it doesn't require to point same MySQL admin database because each node having its own database. 

As you mention above, i am restarting each LAC nodes when import an API.

I want to confirm that Is above case will go fine or Cluster must be configure.

Thank you

shreeram 

You only need to configure Hazelcast if you are using either of this functionality:

- Timer - and only want the timer to run once within the cluster

- MQTT Listener - only one node to process the message.

Otherwise, Hazelcast is not required. 

Starting 5.0, an admin database is no longer required. Each node starts up with a repository (identical) that is bootstrapped and loaded into memory. 

If this is still not clear, please log a support ticket requesting to speak with the product team (referencing this community thread). I'd be happy to discuss this with you offline.

Wai-Yin 

hi chewa03

I really thank you for your explain.  

It's clear now.

shreeram