To:
dnsop@cafax.se
Cc:
mlarson@verisign.com
From:
Tim Seaver <tas@bellsouth.net>
Date:
Tue, 17 Apr 2001 12:08:08 -0400 (EDT)
Sender:
owner-dnsop@cafax.se
Subject:
gTLD server losses
I apologize to VeriSign for mis-stating my gTLD performance results. What I was actually seeing in my DNS tests was a regular 4% loss from the two gTLD servers closest to me, c and g. However, this loss was due to the servers being unresponsive for two 30 minute periods per day during zone reloads rather than being overloaded. With the move of g on April 4, it no longer shows that behavior, and Matt tells me that c will be moving to a new configuration shortly to avoid the reload outages. That's good news, since it should remove some 20% of the server retries I see. Thanks, Tim > Tim, could you please elaborate on your #1 below? I'm not sure if you're > attributing the short TTLs to the {com,net,org} zone contents. But I just > wanted to point out that although all delegations in the {com,net,org} zones > have a 48-hour TTL, that should have little effect on how long a zone's NS > RRset is actually cached. In any BIND name server since 4.9, as well as the > Microsoft DNS server (at least the W2K version--don't have an NT 4 system > handy to test), the NS RRset in the delegated zone overrides the RRset > obtained from the parent. So NS RRsets for subzones of {com,net,org} with > "short" TTLs are an issue with individual zone administrators. > What do you mean by "overload of gTLD servers"? We've sized the > {com,net,org} name servers very carefully and monitor the query volume in > real time. We have lots of headroom and have yet to come close to our > maxmimum. I'd be very interested in any data you've got--I wouldn't like to > think we've missed something. > Thanks, > Matt > -- > Matt Larson <mlarson@verisign.com> > VeriSign Global Registry Services / www.verisign-grs.com > > I'm in the middle of a DNS performance study. Right now, I can > > say with confidence that most DNS performance problems from my > > perspective in the network/DNS topology come from: > > > > 1) Short TTLs on second-level NS records combined with overload of > > gTLD servers, leading to caching server retries for delegation data.