-
1. Re: Production Architecture
jorgemoralespou_2 Apr 22, 2014 3:08 AM (in response to wayne_toh)There is no reference architecture for this :-D
For HA setup, a deployment can be like the following:
- Cluster of nodes for service execution (SwitchYard). Depending on how you distribute services, heterogeneously or homogeneusly, this cluster or clusters can vary.
- Cluster of nodes for rtGov server
- Cluster of nodes for dtGov server.
When I say cluster, I'm not saying 1 or 2 nodes, but more than 1. For dtGov and rtGov 2 can be ok, for service execution, depends on your load. Service instances can be deployed on same boxes (e.g. switchyard1 + switchyard2 on same box), or different boxes, funcitonalities can share boxes (e.g. rtGov Server + switchyard on same box), databases can be on same boxes or different boxes.
A reference architecture is dependent on your application requirements, so there can be guidelines on how to deploy this in an effective HA manner.
Also SwitchYard and rtGov need to be per environment (dev, test, qa, staging, prod, ...) but dtGov can be one for all (also, depending on your architecture).
If you are more specific, we can provide with more guidance.
-
2. Re: Production Architecture
wayne_toh Apr 22, 2014 3:57 AM (in response to jorgemoralespou_2)Below is what i envisaged to be a simple production ha design. The service instance and rtGov are together and accessing a shared database. The dtGov is not required to be highly available and is accessing the SRAMP database. The dtGov is also shared across all environments. Would this be a valid architecture or is there something fundamentally wrong with the design?
-
3. Re: Production Architecture
jorgemoralespou_2 Apr 22, 2014 4:27 AM (in response to wayne_toh)1 of 1 people found this helpfulThis architecture is fine. But as I said, depending upon the requirements for HA, fault tolerance, failover,...
If you have DB instance failover, you need to know the impact on the failing over action time.
Also, need to think if SwitchYard + rtGov are on same EAP instance or different EAP instances. how to load balance the service execution traffic (depending on service heterogeneity or homogeneity), probably using apache as LB with mod_cluster for Http services. Are you using JMS. If so, embedded in same container, in different? Are you using domain or standalone?
But it is the typical setup. It is fine.