[bind10-dev] some other details about differences

[JINMEI Tatuya / 神明達哉]

At Mon, 17 Oct 2011 16:30:05 +0800,
"Likun Zhang" <zlkzhy at gmail.com> wrote:

> Vorner wrote on Saturday, October 15, 2011 4:17 AM
> > Having the maximum number of revisions or maximum number of RRs in the DB
> > seems reasonable. Also, as the RFC suggests, we don't need diffs older than
> > which could be used. If the diffs would be longer than the zone, then there's no
> > need to store them. We don't need that, but we have an upper limit on when
> > they might still be useful. If the diffs table has more than twice as much records
> > as the zone, then it's too large for sure.
> 
> For reference here. RFC1995 also pointed out, if the diff size is bigger than zone, the transfer should fall back to AXFR, so we have to compare the size of diff and zone when we considering the max diff size.

Which part of RFC1995 states that?  I cannot find it.  The only thing
I can find that may be close to this is this:

   Information about older versions should be purged if the total length
   of an IXFR response would be longer than that of an AXFR response.
   Given that the purpose of IXFR is to reduce AXFR overhead, this
   strategy is quite reasonable.
   (Section 5, second paragraph)

But it's about housekeeping, rather than IXFR vs AXFR.  Nevertheless,
of course it doesn't makes sense to keep diffs if the corresponding
IXFR would be larger than the full zone transfer.

BTW, BIND 9 doesn't seem to do this comparison either when choosing
outbound IXFR or AXFR, or when compacting a journal file.

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.