requesting hard data sources on ramifications of verisign wildcard
as already mentioned, fascinating public policy theatre is going on in DC on the verisign wildcard issue, see http://secsac.icann.org/ [all video and even transcripts of both meetings online. go icann.] you are encouraged to read through all of it before making public comments on this issue at nanog. (or, hope springs eternal, to this list.) caida has the following request on behalf of icann's secsac committee [an exceptionally competent group who is approaching this issue with impressive speed, equaniminity, and integrity. http://www.icann.org/committees/security/ i believe we're in good hands here. let's give the process a chance and constructively contribute where we can.] a common theme over the last week is an admitted lack of hard data [rather than lists of theoretical breakages, and anecdotal evidence, and predictions] from the operational community on actual loss of stability in Internet performance or functionality. david from XO gave an outstanding talk on 7 oct, http://www.icann.org/presentations/shairer-secsac-dc-07oct03.ppt but, as with many other providers, he deployed the bind patch within 24 hours so he didn't really have useful hard data to put on the table. i get similar comments from others. ben from harvard also gave some hard alexa data, fwiw http://www.icann.org/presentations/edelman-secsac-dc-15oct03.ppt but from a specific vantage point. we need more of these. icann's secsac committee is in a much stronger position to provide technically sound and equitable guidance if we can provide them with as specific, concrete examples (*hard data*) that indicate extent of various types of breakage. please save the arguments regarding the legitimacy and short notice of this request until you've read the hours of discussion about it that has already occurred among many qualified folks in DC (and in any case the meta-issue still stretches AUP of nanog list). the inconvenient reality is that the secsac committee needs concrete data (imagine), in addition to bulleted lists of things that break, for the policy process to work most effectively here. and nanog is in a position to make a difference. you are hereby encouraged to do so. please send any hard data reflecting observed ramifications on security and stability of Internet infrastructure to secsac-comment@icann.org no hard data will be refused service k
on 10/16/2003 2:26 PM k claffy wrote:
caida has the following request on behalf of icann's secsac committee
a common theme over the last week is an admitted lack of hard data [rather than lists of theoretical breakages, and anecdotal evidence, and predictions] from the operational community on actual loss of stability in Internet performance or functionality.
VeriSign's response to the IAB confirms actual breakage, but dismisses this acknowledgement by claiming that the breakages are insignificantly minor to the Internet community as a whole. This is fodder for the canonical political issue -- the original generic TLDs are public resources, and have operational and management requirements which are significantly more stringent than private TLDs. -- Eric A. Hall http://www.ehsco.com/ Internet Core Protocols http://www.oreilly.com/catalog/coreprot/
k claffy wrote:
... please send any hard data reflecting observed ramifications on security and stability of Internet infrastructure to
secsac-comment@icann.org
no hard data will be refused service
Here's a glimpse of some data for a small ISP (bcc'd to secsac). This mail server was clogging with spam that couldn't be rejected with bad .com and .net incoming addresses, and with bad .com and .net outgoing undeliverable addresses. The server failed (stopped responding to new SMTP requests, and/or crashed) again and again: Sun, Sep 21, 2003 11:52 PM mail.WaterValley.Net 2 minutes, 35 seconds Mon, Sep 22, 2003 00:01 AM mail.WaterValley.Net 4 minutes, 7 seconds Mon, Sep 22, 2003 00:12 AM mail.WaterValley.Net 5 minutes, 48 seconds Mon, Sep 22, 2003 01:18 AM mail.WaterValley.Net 1 minute, 1 second Mon, Sep 22, 2003 04:07 AM mail.WaterValley.Net 5 minutes, 16 seconds Mon, Sep 22, 2003 04:23 AM mail.WaterValley.Net 3 minutes, 3 seconds Mon, Sep 22, 2003 04:33 AM mail.WaterValley.Net 1 minute, 19 seconds Mon, Sep 22, 2003 04:37 AM mail.WaterValley.Net 9 minutes, 4 seconds Mon, Sep 22, 2003 06:47 AM mail.WaterValley.Net 22 minutes, 58 seconds Mon, Sep 22, 2003 07:15 AM mail.WaterValley.Net 6 minutes, 59 seconds ... Mon, Sep 22, 2003 09:53 PM mail.WaterValley.Net 3 minutes, 0 seconds Mon, Sep 22, 2003 10:01 PM mail.WaterValley.Net 5 minutes, 0 seconds Mon, Sep 22, 2003 10:13 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 10:21 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 10:31 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 10:39 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 10:49 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 10:59 PM mail.WaterValley.Net 3 minutes, 1 second Mon, Sep 22, 2003 11:07 PM mail.WaterValley.Net 3 minutes, 2 seconds Mon, Sep 22, 2003 11:17 PM mail.WaterValley.Net 1 minute, 3 seconds Then, A MIRACLE OCCURRED! The problems STOPPED! That miracle was BIND 9.2.3rc3, for which we are eternally grateful. As I posted to NANOG on Tue, 23 Sep 2003 02:35:48 -0400: William Allen Simpson wrote: # Thought I'd mention that I helped setup BIND 9.2.3rc3 on a yellowdog # linux powercomputing machine tonight. It worked. And the mail queues # began clearing out. ... The next downtime (for restoring saved mail queues) was: Wed, Sep 24, 2003 06:39 PM mail.WaterValley.Net 21 minutes, 0 seconds Note the dramatic difference -- from failures several times per hour, to stability for days! I don't know how many others were devastated by the VeriSign wildcards, or whether the differences were as dramatic elsewhere. Hopefully, other ISPs worldwide will step forward. I expect we can come up with more data, but I'll save most of it for the expected future affidavits.... -- William Allen Simpson Key fingerprint = 17 40 5E 67 15 6F 31 26 DD 0D B9 9B 6A 15 2C 32
participants (3)
-
Eric A. Hall -
k claffy -
William Allen Simpson