Groups and SuperGroups

[johnstoni]

I work for a global enterprise that has around a dozen PA 7000 series FWs running between 2 & 6 vSystems each. We have many address groups with more than 500 members, and several "SuperGroups," containing two, or more, of those large groups.

Let's say

address-group	group-1	contains	500	members
address-group	group-2	contains	550	members
address-group	group-3	contains	600	members
address-group	group-4	contains	650	members
address-group	group-5	contains	1000	members
address-group	SuperGroup	contains	address-groups	1, 2, 3, 4, & 5

Which is the best way to write a security policy for these groups:

Rule Name	source-zone	address	dest-zone	address	application	services	action	logging
1	internal	SuperGroup	external	8.8.8.8	DNS	Application-default	permit	whatever
2	internal	group-1 group-2 group-3 group-4 group-5	external	8.8.8.8	DNS	Application-default	permit	whatever
3a	internal	group-1	external	8.8.8.8	DNS	Application-default	permit	whatever
3b	internal	group-2	external	8.8.8.8	DNS	Application-default	permit	whatever
3c	internal	group-3	external	8.8.8.8	DNS	Application-default	permit	whatever
3d	internal	group-4	external	8.8.8.8	DNS	Application-default	permit	whatever
3e	internal	group-5	external	8.8.8.8	DNS	Application-default	permit	whatever

i.e. Is combing, or nesting, groups in to super-groups OK? Is there an impact on performance, or anything else, doing this?

      Is listing groups, rather than having a super-group, a better way to do it?

      Is it better just to have separate rules for each group, like 3a through 3e?

The only difference in the above rules is how the source addresses are configured: in a super-group, a list of groups, or using multiple rules, one address-group per rule.

Looking forward to your replies.

Thanks,

Ian

[LAYER_8]

Ian,

No documented changes in performance here, but firewalls do have maximum rules per device.  

I do have an opinion/preference on this, though. Having the single list items from a troubleshooting/debug perspective makes your life easier. Also from a reporting perspective, if I needed to see all traffic touching a particular rule, having it more granular allows for more focused, quicker investigations. 

Using options 1 or 2 in the above case means I won't be able to find out which devices are touching which rules. If I lump EDLs or regions together, I just see the rule is hit. 

If your network team comes to you with a ticket of some servers having stability and reachability issues, if you have a really pervasive outbound rule, it could muddy the water some. Hope this helps!