ELM 703 DOORS minimum production topology and server specs
Is it possible to run the reporting applications on the same server as the core applications if the production deployment is for a small 5 user setup running on Windows servers?
If so what would the minimum approximate machine specs be?
If the reporting apps must be on a separate server then is it just DCC and LQE that should be on the reporting server or should RS also be on the reporting server?
2 nodes would give 3 servers altogether:
Application
JTS+RM+CCM+GC+LDX+RS
Reporting
DCC+LQE
Database
DB2
What would be typical approximate ballpark minimum specs (cores + RAM) for these 3 machines for a small 5 user deployment with an initially small database?
2 answers
Thanks Davyd.
I also have had evaluation environments running with all apps + DB2 on one server with 16GB ram and 8 logical cores with no problems.
Based on this practical experience I was anticipating that a production deployment would simply involve upgrading the machine specs to something beefier and putting DB2 on its own DBS machine.
But when I ran the interactive deployment guide for 703 it flagged up that a separate server is needed for reporting apps (the 702 interactive guides did not flag this)
The 703 interactive deployment guide then links to the topology page on the wiki. This page lists departmental topology as the minimum production topology
https://jazz.net/wiki/bin/view/Deployment/StandardTopologiesOverview#Departmental_topologies
This topology page says that even for evaluation environments you need a separate server machine for reporting. It says for production deployments the minimum topology is 3 ELM nodes with DCC and LQE each requiring their own separate server machines apart from JTS+RM+CCM etc. (plus DB2 database server making a 4th machine).
Comments
It might sound surprising, but we have a team looking at the deployment architecture, refining the guidance in the deployment wiki. Different applications have different requirements with respect to disk space, disk I/O, RAM, CPU, etc. and this is also changing. It is not always obvious, why the recommendations are the way they are, but they reflect our current understanding and best practices.
I do not think that we have any sizing for 5 users.
If you start with one server, it would be possible to replace that server with a reverse proxy and to move the applications to new servers to scale up. This way you can keep the public URI (e.g. https://myserver.example.com/ ) on the proxy and forward the requests to the servers hosting the applications without having to do a server rename.