Re: [bmwg] Several inconsistencies in current draft-ietf-bmwg-ngfw-performance

["MORTON, ALFRED C (AL)" <[e-mail]>]

Hi Petr,

Thanks for your comments on the NGFW draft.  Please subscribe to bmwg-list to ensure that you see all replies.

A few replies below, [acm]
Al
(as document shepherd)

From: bmwg <[e-mail]> On Behalf Of Redkin Petr
Sent: Wednesday, April 7, 2021 7:01 AM
To: [e-mail]
Subject: [bmwg] Several inconsistencies in current draft-ietf-bmwg-ngfw-performance

Hello,

Please, pay attention at several inconsistencies in current netsecopen draft (saw in 6-7 versions).

  1.  TLS SNI

     *   in part of the document SNI is mandatory
The client endpoint SHOULD send TLS Extension Server Name Indication (SNI) information when opening a security tunnel

     *   in part of the document SNI is optional
For TLS the client MAY use Server Name Indication (SNI).
If using SNI, the server will then perform an SNI name check with the proposed FQDN compared to the domain embedded in the certificate.
[acm]
I want to note our Requirements language, which says that only terms like MUST and SHALL indicate mandatory requirements.  SHOULD is a strong recommendation, and MAY is optional of course, so there are two different requirement levels being used as you point out. I'll let the authors explain their rationale for different requirements levels for SNI (in different tests, perhaps?).


  1.  New connections in sustain phase

     *   From this sentence engineer may assume, that no new connections are established in sustain phase because all clients & connections (it is different concepts) must be established at a ramp up

Sustain phase starts when all required clients (connections) are active and operating at their desired load condition.

     *   But it is not, otherwise we cannot measure CPS metric in sustain phase, for example in traffic mix test
Optional KPIs: TCP Connections Per Second and TLS Handshake Rate

  1.  Client source port

     *   In part of the document source port is variable in range and varies per client connection
The source port range SHOULD be in the range of 1024 - 65535.
The behavior of the client is to sweep through the given server IP space, sequentially generating a recognizable service by the DUT.  Thus, a balanced, mesh between client endpoints and server endpoints will be generated in a client port server port combination.

     *   But it is not clear from this sentence
Each client endpoint performs the same actions as other endpoints, with the difference being the source IP of the client endpoint and the target server IP pool.


Best Regards,
Petr Redkin

Performance research expert,
JSC "InfoTeCS"