There is no specific interdiction to use a single PGSQL DB server. Each app will have it's own DB on that PGSQL instance. Still, I don't recommend it as the load on that server will induce latency for all the products.

We have specified in the docs that for high performance UCMDB instances we need a dedicated DB server with some performance details like SSD drives, the same network, close to the UCMDB server, etc.

They can share the same DB server if you are willing to accept the performance degradation.

Thank you so much for your reply.

We don't have a really huge environment.
We have around 100 operating systems and 250 network devices.
Can it work well along with the load balancers and 2 servers?

350 nodes is not much from a UCMDB perspective but I don't know what DB load will induce SMAX, OBM and NNMi on the same DB server.

You can give it a try if you know that you can request a second DB server for the most demanding DB(s).

What load balancers and what 2 servers?

Hi John,

Thanks for your reply.

We are implementing OpsB + OPTIC + BVD(CDF), SMAX (CDF) & NNMi Classic.  All of these modules will be running in integrated way.

Since our environment is small, in some cases "Micro" as per sizing recommendations of MF but Infra required to host all of these is quite huge.

Purpose of having a centralized Postgres servers is to:

1. reduce Infra size,
2. optimize usage of Db servers by hosting multiple DB instances on active/passive cluster of DB server.
3. or ideally create an active active DB cluster hosting DB's for most of the applications.
4. Reduce day 2 management and patching activities.

Can you please review and guide us how can we optimize the utilization of resources required to host the solution.

Hello Haniyya,

each of those products can run on a separate PGSQL server but collocation of the DBs on a single PGSQL server will bring performance degradation. 

I saw SMAX and CMS instances running acceptable on the same PGSQL server but adding NNMi and OpsB will complicate things a lot.

Hi John,

So can we use one instance for SMAX + CMS and other for NNMi + Ops B by adding some buffer on infra layer to handle performance.

Also can we host OMT on the same instances?

The more PGSQL DBs you then the better. I would keep plane B in place to have additional PGSQL DB servers if needed.

Hello Haniyya,

Within my lab 'test' environment, I have multiple PostgreSQL instances on the same physical server.  Each instance is using a different port for communication.

In my case, I have HCMX/SMAX with OO and CMS each with it's own PostgreSQL instance.  OMT is included as part of the HCMX/SMAX instance.  I do this because each product has different specification/setting requirements for the PostgreSQL instance.  Each instance is then tuned for the corresponding product accordingly.

Because the installation of HCMX/SMAX still uses the classic installation, it is required to be the first installation.  If you add additional suites (Helm), you would add them after HCMX/SMAX.  You can configure them to use different PostgreSQL instances that use different ports.

In my case, I have one root CA certificate and one server certificate for the server for TLS communication and use the same certificate for all of the PostgreSQL instances on the physical server.

Your testing may or may not work as expected depending on a lot of factors.

Regards,

Mark