Is this a recommended practice to move Elastic Search Cluster Outside Of Pega (as we are running into intermittent search index related issues) via building an API and pointing to the same ?
If not recommend,please suggest the reasons or impacts.
Also suggest if a fix exists @ PEGA7.4 version.
Moving search to an independent service is the direction we are going but not available in Pega 7.4 I believe - note that even now it is not completely independent as you do need Pega plugins, but at least no pega engine is required, in general a good idea.
1) Do you recommend implementation of "Separating Elastic Search Cluster" for PRPC7.x version with the following Elasticsearch 2.4.6 version?
2) Irrespective of PEGA and Elasticsearch versions,can this approach induce additional latency because of API calls than a DB or temp storage call? If yes,please provide the tuning recommendations.
2) that would require benchmark testing. I would like to believe running ES as an independent service (much like database itself as a service) should perform better but I am not aware of any published result to support that. Just as importantly, it would be much easier to maintain and troubleshooting with loosely coupled systems.
Finally as a suggestion, try to spend some time with your team evaluating your current 7.x systems and come back with a upgrade plan as you cannot take advantage a lot of powerful features of Pega 8.x with Pega 7.x.