Prepare to host a data collection node

The Splunk App for VMware uses a virtual appliance version of the Data Collection Node (DCN) to collect performance metrics. Splunk distributes this as an Open Virtual Appliance (OVA) file called the Splunk OVA for VMware Metrics.

Splunk configures the DCN with the following default configuration:

• Four cores. 8 vCPUs or 4 vCPUs with two cores with a reservation of 2GHz.
• 12GB memory with a reservation of 1GB.
• 16GB of disk space.

In production, the DCNs communicate with the Collection Configuration dashboard, which runs on the Splunk search head, to retrieve data from vCenter Server. To ensure reliable communication between systems, use static IP addresses and dedicated host names for each DCN. See Collect Data from vCenter Server systems using the VMware API.

Prepare to deploy the DCN

Here's how to prepare to deploy the DCN:

• Identify the vCenter servers and managed ESXi hosts from which you want to collect data.
• Determine the number of DCNs that you want to deploy. Each DCN can collect data from 70 or fewer ESXi hosts, based on the specifications for the 8 core DCN configured with the OVA for VMware with a ratio of 25 to 30 virtual machines per host.
• Each Data Collection Node (DCN) needs at least one CPU core for every 10 hosts from which the DCN is collecting data.
• Estimate the number of CPUs needed for your worker processes with the expectation that a CPU in your deployment will eventually fail at some point. Provision at least one extra CPU to help promote capacity and availability in your deployment.
• Obtain static IP addresses and host names to apply to each of the DCNs.

When you have this information, you can then create the data collection nodes. For more information, see Configure the Splunk OVA for VMware Metrics.

When the add-on is installed on Splunk version 8.0.0 or higher and there are more than 8 worker processes configured for DCNs, the hydra gateway server takes a longer time to respond to the request. For best performance and to avoid job expirations, configure no more than 8 worker processes.