> Mark,
> 
> On Aug 21, 2008, at 5:25 PM, Mark Andrews wrote:
> >     I'm not hoping for the best.  I'm confident that there won't
> >     be major issues.
> ...
> >     Yes change is scary.
> 
> This is, perhaps, a question of perspective.
> 
> The Internet is now a basic infrastructure upon which trillions of  
> dollars (no data, I'm guessing) in transactions and increasingly  
> _national_ economies depend.  I suspect folks who have a higher level  
> of responsibility for that infrastructure, e.g., ICANN/IANA and  
> VeriSign, view mucking with it in fundamental ways somewhat  
> differently than you.
> 
> Given the general level of 'just barely working' on the Internet, it  
> is safe to assume that any change at the root of the DNS will break  
> _something_.  The question is how much and what impact will that  
> breakage have.  ISC, like every software vendor, has frequently  
> released code that has had negative impact on the users of that code,  
> but the impact has pretty much been limited to the users of the code  
> and they have 'opted in' by installing and running that code.  In the  
> case of signing the root, ICANN/IANA and VeriSign will be taking an  
> action that has the potential of impacting pretty much everyone  
> without the folks impacted actually taking any action.  An 'oops' here  
> would likely result in rather strikingly negative repercussions.
> 
> As such, I personally believe these sorts of things require extra  
> ordinary levels of care.  You may feel differently.
> 
> >     Every machine that is setting DO is asserting that it can
> >     handle the responses the roots will generate.
> 
> True.  However, because setting DO is the default, the fact that  
> caching servers set DO says essentially nothing about the  
> infrastructure in which those machines reside.
> 
> While I might agree that it is unlikely that there will be major  
> issues if the root is signed, a lot depends on what you define as  
> 'major'. The impact if you are wrong can be ... substantial.  I  
> personally would prefer to have actual data rather than assertions of  
> confidence.

        Which is why I said look at SE and BR.  Their response
        profile to DO queries will be the same as the roots assuming
        you choose similar key sizes.

        Fragmentation shouldn't be a issue except for DNSKEY queries
        and only clients that are validating are likely to make
        such queries.

        You already have referrals falling into the 512-1200 band
        today, all COM referrals fall into this band.  The addition
        of a DS or NSEC and its signatures just going to change the
        percentage of referrals that fall into this band.  This
        band should not result in fragmentation under IPv4 or IPv6.
        No referrals will exceed 1200 bytes as long as you manage
        your zsk's so that only one signature is generated.

; <<>> DiG 9.3.4-P1 <<>> foo.se @a.ns.se +dnssec +norec
; (2 servers found)
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 1651
;; flags: qr; QUERY: 1, ANSWER: 0, AUTHORITY: 4, ADDITIONAL: 3

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags: do; udp: 4096
;; QUESTION SECTION:
;foo.se.                                IN      A

;; AUTHORITY SECTION:
foo.se.                 86400   IN      NS      ns1.webb.se.
foo.se.                 86400   IN      NS      ns2.webb.se.
foo.se.                 7200    IN      NSEC    foobar.se. NS RRSIG NSEC
foo.se.                 7200    IN      RRSIG   NSEC 5 2 7200 20080828055103 
20080821141241 18048 se. 
ZlrpjdfZ5PfK26Uutc59Adl27W9MxRWtWFofMQGeTDcH79VWbCtAMOKQ 
65Ciouu3wCL1D350mZPZ8/HfQWJCAWVh6ocQyEbRtqjJXsrdDUXg6t/z 
BTO4obAp1j20QPWGKLIjncPC3XHYZXsZSxmwJf6lrqSGuZqYMqUpwBF8 Ng4=

;; ADDITIONAL SECTION:
ns1.webb.se.            86400   IN      A       213.115.221.10
ns2.webb.se.            86400   IN      A       213.115.151.140

;; Query time: 368 msec
;; SERVER: 2a01:3f0:0:301::53#53(2a01:3f0:0:301::53)
;; WHEN: Fri Aug 22 13:01:12 2008
;; MSG SIZE  rcvd: 301

        All NXDOMAIN responses will now fall into the 512-1200 band.
        This will result in TC for [EMAIL PROTECTED] clients.  You already
        handle TC for some PODS referrals today.

; <<>> DiG 9.3.4-P1 <<>> oewyouewy.se @a.ns.se +dnssec +norec
; (2 servers found)
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 63605
;; flags: qr aa; QUERY: 1, ANSWER: 0, AUTHORITY: 6, ADDITIONAL: 1

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags: do; udp: 4096
;; QUESTION SECTION:
;oewyouewy.se.                  IN      A

;; AUTHORITY SECTION:
se.                     7200    IN      SOA     catcher-in-the-rye.nic.se. 
registry-default.nic.se. 2008082200 1800 1800 2419200 7200
se.                     7200    IN      RRSIG   SOA 5 1 172800 20080828013700 
20080822001240 18048 se. 
eqX68S5jhmWDTuddEE0ZuwFTSF5YZ14W06wztpKAQk3caqRdzouDYniA 
Z0XsZhRkQ+AejNxtWbTZ+E67QsuULW6yUjp2g5YAreytYgnYgVniv4xI 
HzUB2xjLnuD/3nF5YkDtVHAjysKxmnW2AfSlDASgujUSt5x0oQHM0EMW 3yE=
se.                     7200    IN      NSEC    0-0.se. NS SOA TXT RRSIG NSEC 
DNSKEY
se.                     7200    IN      RRSIG   NSEC 5 1 7200 20080826144950 
20080820061240 18048 se. 
PfwyUPUmM4MWlMpau2TX+GGoc2YcSOEmA0G9q16o0c+8s7lpCAOHwciL 
CuD2GmOIFxWRItSxcanqT0klDp4DB/l4aRrHDSsir4CMAXS+Xao5mnfB 
KYfQ6EhOV0AK9ymBa3jtZkhadhevd94Eeqc40DFq0+tZZgJ4XwJQ0ff5 4Hg=
oewlite.se.             7200    IN      NSEC    oex.se. NS RRSIG NSEC
oewlite.se.             7200    IN      RRSIG   NSEC 5 2 7200 20080826012458 
20080819141240 18048 se. 
YRXf9tX6PYQRKT2ldQ9U5afNIcqEl4qU2ZW+9h+IqOxAsllp2yT4hJBc 
ItDXIMGbOs/oeLpljscNIVaWMtTqo9aWyAOSQks4C6mQY1vpE+/ob+u2 
F7LVHg/wrRJCx/KfESfH2cGmhjVsy4NpO38Tv5xiaZg6sD9qTGPdqjcK vUA=

;; Query time: 365 msec
;; SERVER: 2a01:3f0:0:301::53#53(2a01:3f0:0:301::53)
;; WHEN: Fri Aug 22 12:48:01 2008
;; MSG SIZE  rcvd: 668

        Mark

> Regards,
> -drc
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: [EMAIL PROTECTED]
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to