Re: [Dots] I-D Action: draft-ietf-dots-telemetry-01.txt

[<[e-mail]>]

Hi Jon, all, 

Please see inline. 

Cheers,
Med

> -----Message d'origine-----
> De : Jon Shallow [mailto:[e-mail]]
> Envoyé : mercredi 5 février 2020 21:24
> À : 'Konda, Tirumaleswar Reddy'; BOUCADAIR Mohamed TGI/OLN;
> [e-mail]
> Objet : RE: [Dots] I-D Action: draft-ietf-dots-telemetry-01.txt
> 
> See inline
> 
> Regards
> 
> Jon
> 
...
> > >
> > > (1) key value range for telemetry: Jon raised this point "These
> keys
> > requires 3
> > > bytes - and telemetry information is going to be difficult to fit
> into a
> > packet.  I
> > > appreciate that comprehension-required Is for numbers less than
> 0x8000 -
> > > perhaps the comprehension-required range is reduced and also has a
> > section
> > > higher up so the total of 0x8000 still stands so less bytes can be
> used
> > here."
> > >
> > >    +----------------------+-------+-------+------------+----------
> -----+
> > >    | Parameter Name       | CBOR  | CBOR  | Change     |
> Specification |
> > >    |                      | Key   | Major | Controller |
> Document(s)   |
> > >    |                      | Value | Type  |            |
> |
> > >    +----------------------+-------+-------+------------+----------
> -----+
> > >    | ietf-dots-signal-cha | 32776 |   5   |    IESG    |
> [RFCXXXX]   |
> > >    | nnel:telemetry       |       |       |            |
> |
> > >
> > > Med: This is a major one. We need to assess the gain, but it is
> possible
> in
> > > theory to update our assignment policies and reassign, e.g., 128-
> 255
> range
> > to
> > > be comprehension-optional (specific for telemetry). This would
> mean that
> > > the telemetry spec will be tagged as updating the base signal
> channel
> > spec.
> > > We need more discussion.
> >
> > Why not change the DOTS telemetry attributes to comprehension-
> required
> > ?

[Med] Telemetry attributes are not mandatory for the signal channel to function. A signal channel message enriched with telemetry data should not exacerbate message failure. 

> > If the server does not understand the DOTS telemetry attributes, it
> will
> > respond with 4.00 error response, and the client can re-send the
> request
> > without the DOTS telemetry attributes.
> 
> Jon> Telemetry is also gated by a different sets of Path-URIs.

[Med] Except when telemetry is also included in an update during a mitigation (S-C or S-C). What I have for this one in my local copy is as follows:

   In order to make use of this feature, DOTS clients MUST establish a
   telemetry setup session with the DOTS server in 'idle' time and MUST
   set the 'server-originated-telemetry' attribute to 'true'.

   DOTS servers MUST NOT include telemetry attributes in mitigation
   status updates sent to DOTS clients for which 'server-originated-
   telemetry' attribute is set to 'false'. 

> However,
> source-prefix attribute comes from draft-ietf-dots-signal-call-home
> and also
> occupies 3 bytes - do we change the source-prefix CBOR Key type to
> comprehension-required?
> 

[Med] Including a source prefix in a signal channel message while not supported by the server must not lead to an error. That attribute cannot be set a comprehension-required. I suggest we don't touch that part.