[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: IPComp rfc2393bis-00
At 09:30 AM 11/24/99 +0900, itojun@iijlab.net wrote:
>>There is a second use - to save the CPU cycles of the
>>CPI-to-algorithm conversion, even when the algorithm is negotiated
>>dynamically. That is the rationale for the note in the CPI definition:
> BTW If you use this optimization, the negotiated IPComp association
> will never get expired if it uses bytecount as expiration signal.
IPComp Association expiration -- by byte-count or time -- is not
defined in rfc2393bis. If an implementation elects to negotiate byte-count
via IKE, it needs to visit the IPCA for each packet, and therefore
a CPI in the upper range may be more adequate.
otoh, the CPI definition also includes:
The CPI in combination with the destination IP address uniquely
identifies the compression algorithm characteristics for the
datagram.
So, implementations may locate the IPCA by IP-addr+CPI, and
do more bookkeeping, such as expiration by byte-count.
avram
References: