[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MEXT] [WGLC] draft-ietf-mext-flow-binding-03

Hi Conny,

Please see my individual comments below:

Conny Larsson wrote:
> 
> Hesham Soliman wrote:
> > Hi Conny,
> >
> > We need the comments now, not after the document is updated :). This
> > is WGLC and we're updating the document as a result of reviews.
> >
> Your draft is only considering the binary flow descriptors and I see no
> reason for this limitation. It has previously been suggested that your
> draft only should handle the transfer and operation of flow descriptors.
> Specific flow descriptor handling should be put in a separate draft.

I do not understand your statement that the WG draft considers only binary flow descriptors. The WG draft does not imply any type of flow descriptor in particular, and reserves sub-option types 17-32 for various types of flow descriptors.

> To be more specific. In case of ASCII format there is no need for the
> FID and FID-PRI fields in Figure 2. They are only used for the binary
> flow descriptors. Besides this all option length fields indicates the
> length of the option in 1-octet unit, not 8-octet units.

I do not understand why you're tying the need for Flow IDentifier and FID Priority to a binary format.

The Flow IDentifier and FID Priority are in intrinsic part of the flow binding protocol extension considered so far by the WG which implies ordering the flow binding rules. 

The need for them do not depend on the format used to describe how to match a specific flow aggregate, it is a design choice that has been made. That a flow binding rule matches a traffic aggregate based on an ASCII or binary description doesn't change the fact that rules have to be ordered.

> The Binding Reference sub-option (section 4.2.1), the Flow Description
> sub-option (section 4.2.2) and the Flow Identification Summary option
> (section 4.3) are binary specific things and should be taken care of by
> that draft, they are not needed for the ASCII format.

Same comment as above:

- The flow description doesn't imply a specific format: 

   Flow Description

         The flow description corresponding to the type indicated by the
         Sub-opt Type field.  Flow description is out of scope of this
         document.

   The following values are reserved for the sub-option Type values are
   defined for Flow Description:

      17-32 reserved for Flow Description formats.

- The flow identification summary doesn't imply a specific format:

   The Flow Identification Summary Mobility option includes one or more
   flow identifiers (FIDs) for the purpose of refreshing their state.

> Making these changes, and perhaps some minor modifications related to
> these changes, would make the draft look much better. I believe it's
> doable.

At this point in the lifecycle of a WG document (i.e., the WG Last Call) I don't think we want to change the basics of the flow binding mechanisms, that is, that packets are matched against an ordered ruleset.

--julien