Dear authors,

From the title "Framework for Interface to Network Security Functions”, I 
assume the focus of this draft addresses the controller NSF facing interface. 
Based on that assumption, I have few comments and questions. I would appreciate 
any response or a sit down in Berlin to understand it better.

  1.  Section 3.2, (3) Rule Provisioning —> Do you mean Rules as defined in 
section 9.2?
  2.  Section 3.3 —> Registration Interface —> Does it mean the 
capabilities/functionality a particular instance of NSF/vNSF support? If it is 
true, then in the figure 1, why does this interface need to be treated 
different than other NSF facing interface. I don’t understand the need for mgmt 
system just for this interface.
  3.  Section 4 —> Are we talking about a need for RBAC model here in this 
section?
  4.  Section 5 —> This comment is in general though but since this section 
recommends that NSF facing interface must be based on  Flow/packet information, 
it begs a question whether we leave any big hole, that can not be covered and 
would require customer to deploy multiple solutions. This may not be the right 
place since I think, this is called out in the I2NSF charter. I wanted to bring 
it up though.
  5.  Section 6.1 —> I assume, we don’t want to state a particular 
authentication scheme but I think, it would be good to mention that I2NSF 
controller must support multi-tenant RBAC mechanism (something like this).
  6.  Section 6.2 —> There is some mention of authentication/authorization 
between controller and NSF but only in the open environment. Does it not make 
sense for NSF to operate in one mode (always authenticate)?
  7.  Section 6.3 —> In my opinion, whether it isa  vNSF or pNSF , we could 
have cluster or active/backup deployment in either scenario and security should 
not worry care about the implementation of NSF itself. The security controller 
should interact with the configuration entity on NSF and let it handle 
distributed nature of underlying NSF. This would be a cleaner approach in my 
opinion.
  8.  Section 7.1 —> It looks like user-intent based policy definition. Is my 
understanding correct?
  9.  Section 8 & 9 —> Are we saying that Capability is for Client side and 
Registration is for NSF side?  I am not able to understand the distinction 
between Capability and Registration interface. Can’t we achieve this with one 
interface (capability/discovery) ?

Thanks & Regards,
Rakesh
_______________________________________________
I2nsf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2nsf

Reply via email to