VM Resource Spec for Dedicated Deployment Server- Is there a limit of number of forwarders?

[nicolasbagsarpa]

Is there a limit of number of forwarders that a single dedicated deployment server can connect with?

We're trying to check what will be the vm specification (cpu/memory/disk/etc) that a dedicated deployment server can still managed like for 5k forwarders, 10k, or 30k forwarders. If its possible or not for one server.

I only found documentation about the estimated time-to-deploy calculations, but nothing about the vm-spec vs number-of-forwarders calculation.

[gcusello]

Hi @nicolasbagsarpa ,

there isn't a reference hardware for the Deployment Server.

The only indication is that is a standard Splunk server, so at least 12 CPUs and 12 GB RAM.

In general, you can use also less resources (8+8).

In y experience I never found probles also scaling the number of managed clients.

Also because for this role you don't need a physical server but a virtual one, so you can scalate if the load is too high, but i don't think that it will happen.

You can also use a two tier architecture for the DS, but it isn't an immediate configuration architecture and it requires a Splunk Architect with experience in large architectures.

Ciao.

Giuseppe

[isoutamo]

Hi

here is what splunk said on doc

•  https://docs.splunk.com/Documentation/Splunk/latest/Updating/Calculatedeploymentserverperformance
• https://docs.splunk.com/Documentation/Splunk/latest/Updating/Planadeployment

With VMware or other virtualised platforms I like to start with smaller (e.g. 4vCPU, 4+GB men) and monitoring its resource usage. Before it run out of resources just add more vCPUs, mem, disk etc. When limits of one node will rise then add another node and LB before those node and configure those to support several DSs. There are some answers about it and also conf presentations. To do it easier later, you should use DNS name for your DS server which you can later point to LB instead of individual server.

r. Ismo