[dns-operations] BIND 9.8.0 and TTL banding

Michael Sinatra michael at rancid.berkeley.edu
Fri Feb 25 00:53:29 UTC 2011


On 2/24/11 3:59 PM, Michael Graff wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> BIND 9 has had a controversial feature called "RTT banding" since 9.5,
> where the resolver would choose from more than one server in an NS set
> if the RTT times were "close," where close was within 128 ms.
>
> We've decided to remove this.  This will cause some servers to get
> slightly different query load patterns, but we believe for the
> resolver-side, the net effect is a positive one overall.

I take it that this will cause BIND 9.8.0's behavior to revert to the 
simple RTT measurement method that was used before RTT Banding was 
introduced?  This as opposed to a continuously random auth server 
selection....

michael



More information about the dns-operations mailing list