Experts,
My customer is trying to redesign ISE on VM’s correctly so that it works in the most efficient manner on a shared storage infrastructure. The processor, memory and disk space requirements are clear, but they need help with optimizing the shared storage network. The customer would prefer access to a Tested Reference Configuration or Preferred Architecture that they can mimic if it is available. In lieu of Preferred/Tested Reference Design, the customer just wants help zooming in on the best possible storage design for ISE. They seem to have every type of shared storage available to them, FC, FCoE, NFS and they can use HDD’s or SSDs. They also have various performance controllers they can use. They want to understand what would perform the best.
Here’s what I have for the environment:
ISE
Is licensed for up to 25K concurrent connections
Has 10 nodes deployed and enough licenses to go to 20 nodes
The nodes are deployed across the 2 large campuses
Primary Admin node and primary reporting node at Campus 1
Backup Admin and reporting nodes are at Campus 2
3 Policy Nodes on Campus 1 and 3 on Campus 2
There is a 10Gig intercampus link between them
In lieu of a Preferred Architecture for storage can someone answer these questions so the customer's storage experts can figure out the most efficient storage design:
what tests are being used to measure I/O throughput?
How much data is being transferred during the test and does this represent typical application reads/writes?
Is it an iometer test?
is there another test that is being used?
What is the block size of the data?
What are typical types and sizes of the reads and writes that the applications will perform?
Thanks!
Paul Merlitti
