Differential updates in Multi GB zone files.

Maren S. Leizaola leizaola at sarnic.hk.com
Tue May 30 12:15:09 UTC 2000


Hi,
	How does one manage zone updates of when having multiple DNS
servers which hold a multi-GB zone file.

	Are there any extensions or scripts out there to do this? How do
the root servers move the zone file around? using zone x-fer? Or do you
gzip it and ftp it over?

	Maybe I am reinventing the wheel but it would be neat that rather
than just transfer a huge file from one master to n slaves which are in
geographically disperse locations/networks, to actually add an extension
the the zone x-fer protocol and make it a bit more intelligent.

	When updating a zone file, it would be good if the server would
keep the differences (insertions/deletions) between the zone file of the
previous serial number and the current one by doing a comparision of
serial numbers or keep updates in a journal file.

	When a slave is sent a notify or a slave does a AXFR, the slave
would would get the changes not the whole zone file. BIND could keep a
hash of the last zone it has in memory when updates are done, slave and
masters could exchange hashes to ensure that they both hold the same data.
If the hash fails, then it does a old fashioned AXFR.

	Maybe my idea is not too polished but I am sure someone can come
up with something along these lines to solve this problem.

A solution which I am not sure if will work....

	I see that BIND 9.0 has some database extensions to host the zone
files in an SQL back end. The sql backend could be used to do the zone
replication, which is neat, good bye zone updates, here come real time DNS
changes. How well would an SQL back end handle huge loads compared to the
current bind? Can someone teach me a bit on this?

Regards,
Maren.




More information about the bind-users mailing list