[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Packet-by-packet compression within IPSec
At 06:25 PM 11/21/96 -0500, Stephen Kent wrote:
>Bob,
>
> I have no objections to your proposal, and it certainly can be
>easily integrated into the newly revised ESP spec. We also can allow for
>an optional, variable length field in front of the payload, that contains
>any per-packet data needed for a specific algorithm. This is easy to
>include in the ESP spec, with SA negotiation determining the presence or
>absence of the field and its size.
Thanks. FYI, the only per-packet data I am imagining for our particular
proposal would be a single byte to contain a compressed/uncompressed bit.
This is needed to handle the case where the source data expands and you
want to instead send the original uncompressed data. In this case, you
would set the bit saying that the particular packet is uncompressed, even
though the SPI specifies that compression is an active function for this
channel.
>The only issue,is how we deal with the
>possible requirement to support any specific compression protocols.
When you say, "...possible requirement to support any specific compression
protocols.", I'm not sure I understand the issue. Do you mean the minimum
or mandatory support levels for compression in IPSec/ESP? If so, I would
expect that its optional nature precludes any such requirement. If you mean
support for specific compression algorithms, I would suggest that we treat
that issue in a similar manner as was done with PPP, where any number
algorithms can be supported as long as there is a draft/standard document
to describe how it is done in the context of IPSec/ESP. Since the specific
compression algorithm that we will be proposing will be based on the LZS
algorithm, it will ultimately result in an informational RFC (which is what
happened in the PPP case for LZS as well).
>From an interoperability perspective, we need to address this aspect of the
>standards, and that, I believe, is the basis for concern in terms of
>delaying deployment of this technology.
Are you refering to the requirement for two interoperable implementations
of compressed ESP before the document can be moved to the Draft Standard
stage? Having done some recent homework on the IETF standards process, I'm
guessing this is the concern. If that is indeed the case, I believe that we
could easily achieve such a goal.
-Bob