Forum Discussion
You would be looking at either 4 cores (R2600) or 8 cores (R2800). Please see the following: https://clouddocs.f5.com/training/community/rseries-training/html/rseries_multitenancy.html. Sadly, this platform only support 1 tenant, so presume you would be running HA across 2 devices with only one device responding to both LTM and GTM (DNS) connections. Of course, you can have 2 traffic groups and use another traffic group with a DNS listener to run DNS active active across both units while LTM is active/standby. Usually best practice is to have smaller dedicated appliances, or carved out vCMP guests (2 cores, if not handling Internet requests) dedicated for GTM and another instance for LTM with more cores. However, this is not an option on the R2xxx platform. Depending on the traffic, and number of LTM VS, that should steer you in the direction of which model to get. Keep in mind, you can always convert via license an R2600 to a R2800, so you can start small and grow as needed. In terms of DNS performance, you are looking generally at 125K DNS requests per core on the F5 platforms if DDoS mitigation is a concern (that is why I am usually hesitant to use other products like Infoblox for external DNS as they would melt).
- zamroni777Nacreous
DNS module and DNS traffic itself is quite negligible, even for VE
https://my.f5.com/manage/s/article/K13114858and moreover client also will also caches dns response (for duration defined in f5 dns config).
so basically you only need to consider load of LTM.
you can use basic calculation from here:
F5 BIG-IP Sizing Requirement | DevCentral