AltostratusWe have a DC and DR and till now wide IP was configured for global availability to point to HQ servers. We have 20 plus WIPS, 5 being public and rest internal. Now management needs active-active setup. So thinking of using round robin with persistence. So will using the persistence with default values make the ideal choice? Most of users are internal users, so only 3 LDNS.
CumulonimbusFor internal DNS you have more control and can tweak the DNS and GTM configuration to best fit your needs. Assuming the internal DNS is not using any sort of Anycast or other method of load balancing client DNS requests to your three DNS servers, then you can start out with the default persistence setting for your internal WIPs but you might want to update the Persistence TTL.
For your external WIPs it really depends, the general public landscape for DNS on the internet is a lot different from several years ago with a lot of companies/ISPs using Anycast and local load balancing for client DNS requests.
This means it is possible, and likely, client DNS queries will go to a different recursive DNS server and so the IP seen by the GTMs will differ and brake your persistence configuration.
To combat this a little I would change the default Persist CIDR (IPv4) values down from 32 for your public WIPs. You will likely need to monitor and see how many, if any, clients are getting flipped between your data centres, I have had customers working fine with CIDR of /32 while others still getting clients flipping on a /21 it depends a lot on the application, time users spend within a single session and where the users are connecting from.
As an alternative you could look at Topology where you could direct users from different regions to a specific Data Centre with the other being a backup, in effect creating mini active-passive solutions based on the client location, good if you have client from a wide range of areas.
Hope this helps :)