Re: [DNSOP] Fw: New Version Notification for draft-yao-dnsop-root-cache-00.txt
Bob Haroldwrote: > > Reading these various ideas brings up a question in my mind. If a server > queries for the SOA of a zone and the serial number has not changed, can it > then assume that all of the entries in its cache for that zone should still > be valid now, and for the their original TTL value starting now? As well as what Robert said, SOA serial numbers are (basically) only relevant for the zone transfer protocol; if a zone is dynamically synthesized you can expect to get different answers for its contents without its serial number necessarily changing. Tony. -- f.anthony.n.finch http://dotat.at/ Shannon, Rockall: West or southwest, backing south, 6 to gale 8, increasing severe gale 9 at times, then decreasing 4 or 5 later. Very rough or high, occasionally very high. Occasional rain or showers. Good, occasionally poor. ___ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
Re: [DNSOP] Fw: New Version Notification for draft-yao-dnsop-root-cache-00.txt
On Tue, Sep 29, 2015 at 5:20 AM, Shane Kerrwrote: > Jiankang Yao, > > I think a simpler approach that works in general is the "HAMMER" > approach proposed by Warren Kumari, Roy Arends, and Suzanne Woolf a > couple of years ago: > > https://tools.ietf.org/html/draft-wkumari-dnsop-hammer > > Basically the idea is that if a query is made for a RRSET that is near > expiration from the cache, then the resolver will answer normally but > will also try to refresh the TTL by performing another query. > > Note that Unbound already implements something like this today, with > the "prefetch" option: > > https://unbound.net/documentation/unbound.conf.html > > BIND 9 does as well, with "prefetch": > > https://deepthought.isc.org/article/AA-01122/0/Early-refresh-of-cache > > The "HAMMER" approach works for all domains, not just the root zone, > and doesn't require any separate cache, or indeed any additional state > at all. > > The approach you propose will have some small advantage if someone > queries for an entry in the root zone that is not in cache. However > given the long TTL of root zone entries, such a query will be rare so > the benefit is quite small. > > Cheers, > > -- > Shane ... > > > Jiankang Yao > > > > From: internet-drafts > > Date: 2015-09-29 12:20 > > To: XiaoDong Lee; Jiankang Yao; Xiaodong Li; Jiankang Yao; Ning Kong; > Ning Kong > > Subject: New Version Notification for draft-yao-dnsop-root-cache-00.txt > > > > A new version of I-D, draft-yao-dnsop-root-cache-00.txt > > has been successfully submitted by Jiankang Yao and posted to the > > IETF repository. > > > > Name: draft-yao-dnsop-root-cache > > Revision: 00 > > Title: Decreasing Fetch time of Root Data by Improving the Mechanism of > Root Data Cacheing > > Document date: 2015-09-28 > > Group: Individual Submission > > Pages: 10 > > URL: > https://www.ietf.org/internet-drafts/draft-yao-dnsop-root-cache-00.txt > > Status: > https://datatracker.ietf.org/doc/draft-yao-dnsop-root-cache/ > > Htmlized: > https://tools.ietf.org/html/draft-yao-dnsop-root-cache-00 > > > > > > Abstract: > >Many DNS recursive resolvers have long round trip times to the DNS > >root server. It has been an obstacle to increse the performance of > >DNS query. In order to decrease fetch time of DNS root data, this > >document proposes a new mechanism by improving the mechanism of root > >data cacheing. > Reading these various ideas brings up a question in my mind. If a server queries for the SOA of a zone and the serial number has not changed, can it then assume that all of the entries in its cache for that zone should still be valid now, and for the their original TTL value starting now? If the values had changed, wouldn't the serial # also change? Seems like I must be missing something here. -- Bob Harold ___ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
Re: [DNSOP] Fw: New Version Notification for draft-yao-dnsop-root-cache-00.txt
Bob Harold wrote: > Reading these various ideas brings up a question in my mind. If a server > queries for the SOA of a zone and the serial number has not changed, can it > then assume that all of the entries in its cache for that zone should still > be valid now, and for the their original TTL value starting now? If the > values had changed, wouldn't the serial # also change? Seems like I must > be missing something here. No inferences like that can be drawn based on the SOA SERIAL field, because the serial number may have wrapped around to the same value that was observed previously. (Even if the time between queries is very small, there is still a finite window of time during which the zone publisher can fit as many zone updates into as needed -- at least conceptually.) -- Robert Edmonds ___ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
Re: [DNSOP] Fw: New Version Notification for draft-yao-dnsop-root-cache-00.txt
Jiankang Yao, I think a simpler approach that works in general is the "HAMMER" approach proposed by Warren Kumari, Roy Arends, and Suzanne Woolf a couple of years ago: https://tools.ietf.org/html/draft-wkumari-dnsop-hammer Basically the idea is that if a query is made for a RRSET that is near expiration from the cache, then the resolver will answer normally but will also try to refresh the TTL by performing another query. Note that Unbound already implements something like this today, with the "prefetch" option: https://unbound.net/documentation/unbound.conf.html BIND 9 does as well, with "prefetch": https://deepthought.isc.org/article/AA-01122/0/Early-refresh-of-cache The "HAMMER" approach works for all domains, not just the root zone, and doesn't require any separate cache, or indeed any additional state at all. The approach you propose will have some small advantage if someone queries for an entry in the root zone that is not in cache. However given the long TTL of root zone entries, such a query will be rare so the benefit is quite small. Cheers, -- Shane On Tue, 29 Sep 2015 12:28:06 +0800 "Jiankang Yao"wrote: > > Dear all, > > I submit a draft about Decreasing Fetch time of DNS Root Data. > >Many DNS recursive resolvers have long round trip times to the DNS >root server. It has been an obstacle to increse the performance of >DNS query. In order to decrease fetch time of DNS root data, this >document proposes a new mechanism by improving the mechanism of root >data cacheing. > >Pls kindly help to review it and give the comments. > > I would also like to apply 10 minutes slot to introduce this idea in > the next IETF meeting > > thanks a lot. > > > > > Jiankang Yao > > From: internet-drafts > Date: 2015-09-29 12:20 > To: XiaoDong Lee; Jiankang Yao; Xiaodong Li; Jiankang Yao; Ning Kong; Ning > Kong > Subject: New Version Notification for draft-yao-dnsop-root-cache-00.txt > > A new version of I-D, draft-yao-dnsop-root-cache-00.txt > has been successfully submitted by Jiankang Yao and posted to the > IETF repository. > > Name: draft-yao-dnsop-root-cache > Revision: 00 > Title: Decreasing Fetch time of Root Data by Improving the Mechanism of Root > Data Cacheing > Document date: 2015-09-28 > Group: Individual Submission > Pages: 10 > URL: > https://www.ietf.org/internet-drafts/draft-yao-dnsop-root-cache-00.txt > Status: https://datatracker.ietf.org/doc/draft-yao-dnsop-root-cache/ > Htmlized: https://tools.ietf.org/html/draft-yao-dnsop-root-cache-00 > > > Abstract: >Many DNS recursive resolvers have long round trip times to the DNS >root server. It has been an obstacle to increse the performance of >DNS query. In order to decrease fetch time of DNS root data, this >document proposes a new mechanism by improving the mechanism of root >data cacheing. > > > > > > Please note that it may take a couple of minutes from the time of submission > until the htmlized version and diff are available at tools.ietf.org. > > The IETF Secretariat ___ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
[DNSOP] Fw: New Version Notification for draft-yao-dnsop-root-cache-00.txt
Dear all, I submit a draft about Decreasing Fetch time of DNS Root Data. Many DNS recursive resolvers have long round trip times to the DNS root server. It has been an obstacle to increse the performance of DNS query. In order to decrease fetch time of DNS root data, this document proposes a new mechanism by improving the mechanism of root data cacheing. Pls kindly help to review it and give the comments. I would also like to apply 10 minutes slot to introduce this idea in the next IETF meeting thanks a lot. Jiankang Yao From: internet-drafts Date: 2015-09-29 12:20 To: XiaoDong Lee; Jiankang Yao; Xiaodong Li; Jiankang Yao; Ning Kong; Ning Kong Subject: New Version Notification for draft-yao-dnsop-root-cache-00.txt A new version of I-D, draft-yao-dnsop-root-cache-00.txt has been successfully submitted by Jiankang Yao and posted to the IETF repository. Name: draft-yao-dnsop-root-cache Revision: 00 Title: Decreasing Fetch time of Root Data by Improving the Mechanism of Root Data Cacheing Document date: 2015-09-28 Group: Individual Submission Pages: 10 URL: https://www.ietf.org/internet-drafts/draft-yao-dnsop-root-cache-00.txt Status: https://datatracker.ietf.org/doc/draft-yao-dnsop-root-cache/ Htmlized: https://tools.ietf.org/html/draft-yao-dnsop-root-cache-00 Abstract: Many DNS recursive resolvers have long round trip times to the DNS root server. It has been an obstacle to increse the performance of DNS query. In order to decrease fetch time of DNS root data, this document proposes a new mechanism by improving the mechanism of root data cacheing. Please note that it may take a couple of minutes from the time of submission until the htmlized version and diff are available at tools.ietf.org. The IETF Secretariat___ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop