ISE Scale

[dazza_johnson]

Is it standard practise to have the bigger SNS-3595s as the PAN/MnT and the smaller SNS-3515s as the PSNs in a distributed deployment. I am 95% sure it is, but want to make sure.

[Charlie Moreton]

Standard practice is SNS-3595 throughout.  All of the published numbers for ISE Distributed Deployments are written as every node is a or sized as a 3595.

For the Medium-Sized deployments, it is also assumes that the platform for the PSN is the same as the Admin/MnT.

Mixing them will not yield the same numbers as have been published.

ISE Deployment and Scaling

[Craig Hyps · Accepted solution]

"All of the published numbers for ISE Distributed Deployments are written as every node is a or sized as a 3595."

This is not a correct statement.  We perform Performance and Scale testing based on the various deployment models listed in above link.  These numbers are based on PnS testing for the various platforms including 3415, 3495, 3515, and 3595.    It is correct that we always assume a large appliance (3495 or 3595) for all deployments where each persona runs on a dedicated node.  The per-PSN scaling figures are also rated based on specific platform.  Even our multi-node/multi-service scale testing where we represent a full deployment includes options with mixed nodes.

Let me know if seeing anything in the documentation that would imply that our scaling numbers are predicated solely on 3595.

In response to original question, I assume question is specific to the middle tier deployment model.  If standalone or fully dedicated model, then specific guidance is provided.  For deployments middle-tier deployments with collocated PAN+MnT and distributed nodes, and asking if you should use 3595 for PAN+MnT versus PSNs, then I recommend PAN+MnT node.  The total capacity of deployment is gated based on the PAN/MNT node and MnT can best leverage the faster and higher-capacity disk system in 3595. It also sets you up for greater scale and expansion to fully dedicated deployment.

/Craig