ISE: setting up the pxGrid nodes

[filip.johannes.hansen]

Got customer where we are doing a greenfield installation of ISE with a mix of HW and VM. The setup wil be integrated with sd-acces (dna) and it will be run a in multidomain setup (aci pahse 3 when that become availble)

atfer reading the recommedation documentation that you should be using  seperate  pxGrid personas in a large deployement that has been decided upon. And the customer will be using  2 VM 's for this

My question is this how do you spec the pxgrid personas? they are not goingto be using more than max 10 subscribers are some of the limitations i can find (25 I think the number was for the smalle one? ) but how do you spec it according to the HW resources it will use? do yo use the small/medium/large setup on the common template? or  is there a seperate HW (vm) spec when you are runing pxGrids by themselves seperate from the pan node?

any feedback wil be much appreciated 

or if anyone has good documentation on this they can point me to 

[Greg Gibbs · Accepted solution]

Apart from the guidelines around maximums listed in the https://cs.co/ise-scale doc, it's difficult to provide detailed scaling for pxGrid. It's highly transactional and there are multiple variables involved.

The nodes still have to meet the hardware spec requirements for Small/Medium/Large and would use the same resources as a PSN. If DNAC is currently your only subscriber, you could start with dedicated PXG nodes based on the 3615 (with the PSN OVA). If your subscribers increase in future and the node CPU/Mem resources are getting hammered, you could update the VM resources to meet the 3655 platform.

[filip.johannes.hansen]

Thanks did some more digging as well and came to the same conclusion. Appreciate you taking the time and effort to answer