- NANOG - lists.nanog.org

Traffic statistics from major ISPs
by Sean Donelan 02 Jan '00

02 Jan '00

In the voice world, AT&T, SBC, Bell Atlantic and others have released traffic figures about how many voice calls they handled around New Year's. I have not seen similar data from any major ISPs about traffic volume around New Year's. I know about third-party sites such as Keynote, and MIDS; but they don't measure traffic volume. I'm looking for basic data such as ANS used to publish for the NSFNET. The closest I've found is Abovenet's MRTG pages, but it is spread out over different links and its hard to aggregate the numbers without double counting, or undercounting. I know several major ISPs collect the data internally, but haven't released it like the voice side of the house.

2 1

RE: mail does bounce (was: Customers down?)
by Sean Donelan 02 Jan '00

02 Jan '00

On Sat, 01 January 2000, "Roeland M.J. Meyer" wrote: > > I agree with Greg Woods - if a domain is authoritatively non-existent, > > I'd expect a sane mailer to bounce the message. > > I'm glad that you don't write code, especially sendmail code. Trade-offs. Would you rather be notified immediately if you mistype an address, or wait four days before getting an undeliverable bounce. If the name exists, but the mail host is unreachable try the next MX record or if no more alternates, queue the mail. If the name does not exist, return mail to sender. If you can't figure out whether the name exists or not, do what? The RFCs just say a mailer is expected to do something "reasonable" in reponse to a domain server error. But RFC 974 doesn't specify what. An otherwise compliant mailer can queue the mail or bounce it, and still be considered compliant. If a site wants to avoid this unspecified behaivor, a wise site would try to assure availability of their DNS servers perhaps by using network diversity so a single failure won't take down all its servers. I remember setting up a connection last year with a major unnamed provider, and pointed out a similar issue. The other provider replied they had an extremely reliable network, and there will never be any problem with their DNS servers. Less than a month later, the extremely reliable network barfed, with the expected consequences.

1 0

Re: Well, MOST places got it right....
by Peter Murray 02 Jan '00

02 Jan '00

--On Sat, Jan 1, 2000 12:32 PM -0500 David Lesher <wb8foz(a)nrk.com> wrote: > http://www.cwru.edu We've been planning this for months: finding an "old fashion" font, scanning pictures from the university's archives and yearbooks, etc. The response that we've gotten to our webmaster account so far has been priceless. "Share and enjoy!" Peter -- Peter Murray, Library Systems Manager pem(a)po.cwru.edu Digital Media Services http://www.cwru.edu/home/pem.html Case Western Reserve University, Cleveland, Ohio W:216-368-8834

1 0

Y2K Funny
by Tim Wolfe 02 Jan '00

02 Jan '00

Well I'm sure I'm not the only one who heard this but NBC news just reported a story that the USNO website listed the date as 01 Jan 19100. So much for the most accurate clock in the world, heh. -- Tim -------------------------------------------------- * Timothy M. Wolfe, Chief Network Engineer * * ClipperNet Corporation / It's a wireless world * * tim(a)clipper.net 800.338.2629 x 402 * * Sufficient for today = Inadequate for tomorrow * --------------------------------------------------

4 3

Re: Customers down?
by Sean Donelan 01 Jan '00

01 Jan '00

On Fri, 31 December 1999, jlewis(a)lewis.org wrote: > I guess our customers aren't that silly. I did notice some big sites down > though. I tried to check a price with Tech Data earlier this evening and > found that www.techdata.com didn't exist. Traceroute to their > nameserver(s) > > techdata.com. 2D IN NS NS1.techdata.com. > techdata.com. 2D IN NS NS2.techdata.com. > NS1.techdata.com. 2D IN A 169.153.202.40 > NS2.techdata.com. 2D IN A 169.153.202.41 > > ends in a CW routing loop suggesting TD shut off their connection for the > evening. Putting your only name servers on the same subnet (unless Tech Data has VERY small subnets) isn't very smart no matter what day it is. Too many things can take down your one network. Instead of queueing mail, many mail programs will generate permanent errors and immediate bounces. Gun, Foot, Self.

5 4

Y2K
by Paul Ferguson 01 Jan '00

01 Jan '00

Well, here's to a major non-event. Cheers. Thank goodness. I chalk it up to a lot of hard work over the course of the past 5 years by a lot of people (some of coming very recently), and a smidgeon (well, maybe more than a smidgeon) of media hype. Again. In any event, let's hope everyone's Y2K is no more eventful than the last year. plop, - paul

4 4

Re: Curious thing in a Cisco router
by Sean Donelan 01 Jan '00

01 Jan '00

At 10:17 PM -0200 12/31/99, Rubens Kuhl Jr. wrote: >While monitoring GMT Y2K progression on a Cisco router, something curious >showed up: [...] >Notice the dot before the time; it was not appearing before, and even on the >first sample after GMT Y2K-rollover (local time is GMT -0200). It now shows >up on every 'show clock'. This doesn't seem to be a problem with cisco, but if you are synchronizing to certain NTP servers (i.e. usno or nist) they currently have reachability problems. So your cisco reports a "." that your ntp is unsynchronized. I haven't found out if the government networks are deliberately throttling traffic or if the links to the time servers are just congested. I have no idea why it started at almost exactly midnight UTC.

2 1

Re: Y2K bgp announcements
by Sean Donelan 01 Jan '00

01 Jan '00

On Sat, 01 January 2000, Ben Buxton wrote: > Looks like numerous networks went off the air over the year rollover > due to fear and the like, and this caused a significant drop in > the number of announcements: The precise numbers will vary depending where your BGP view is in the worldwide Internet. You are correct. There was a steady decline in both number of prefixes and number of asns in the 24 hours before Midnight. The global routing table shrunk by 1000 routes before the hour before Midnight EST. In that hour, another 600 routes withdrew. The routes remained stable at that level for 5 hours, and then began to climb back to normal levels. About 30 ASNs vanished prior to the rollover. >From my view, the decline actually started on December 24. But it isn't really noticable until the last 24 hours. Since the graph of practically anything about Internet always shows an increase, a decrease is an interesting change.

1 0

Correction: www.time.gov didn't make it?
by Sean Donelan 01 Jan '00

01 Jan '00

The report should read NIST was overwhelmed with traffic last night, and the ntp servers were so busy servicing requests they didn't have time to check their clocks. They set a "yellow flag" to indicate the time may be off. They still have a lot of traffic, and may continue to have problems.

1 0

Y2K bgp announcements
by Ben Buxton 01 Jan '00

01 Jan '00

Looks like numerous networks went off the air over the year rollover due to fear and the like, and this caused a significant drop in the number of announcements: http://www.cactii.net/stats/bgp.html This was a drop of about 1 or 2 thousand routes over the 24 hour period. :) -- BB - Random Network Person

1 0