[This was started last month. been a little busy. unsuprisingly I
only had to *add* an incident and it still works.]
- this month, another knee was at 150k [Dec 4th] and similarly
garbled results came out. Again, no response.
...in this one year we've seen the shape of the climb return to the
curve characterized by two years 99-01. Going for e? I'm not quite
sure what the current point of the report is if no-one responds to
even it breaking.
Knee? Shape? Curve? Are you reading the same CIDR report
that I see here every Friday? The report that I see is
basically a dump of raw data. Perhaps the author needs
to remember the distinction between data and information
and make the CIDR report into something that people
*WANT* to read. This posting of yours contained far more
information than any CIDR report.
Those believing otherwise are encouraged to send real, hard data.
There is no meaningful data I can find since the Bellovin/Bush/
Griffin/Rexford 2001 paper.
I don't know why people like to post cryptic references
to documents or meetings etc. Perhaps there is an implicit
desire to hide it from outsiders who are not part of the
secret inner circle? Perhaps this type of behavior is at
the root of the growing problems, i.e. clue is not being
spread around because people are too cliquish in the way
that they present this info. I'm not talking about NANOG
meetings here, just the list, which arguably reaches more
people than the meetings.
Now, on to Bellovin et al. Janet Rexford wrote this paper
on filtering http://www.research.att.com/~jrex/papers/filter.pdf
This was presented at NANOG but the NANOG site here
http://www.nanog.org/mtg-0105/prefix.html points to
Janet's site here http://www.research.att.com/~jrex/nanog/lost.html
(note the word LOST in the URL) which points to Randy's
slides here http://psg.com/~randy/010521.nanog/
unfortunately those appear to be lost...
But there is some video from IETF 51 here
http://videolab.uoregon.edu/events/ietf/ietf51.html
which might be the same stuff. Bush talks on
Routing Issues.
One would think that if the problems noted in 2001
are not being solved, then perhaps a review of this
material might prove more fruitful than more
studies of the data. According to the plot here
http://www.cidr-report.org/ the average announcements
per origin AS has actually taken a turn for the better.
And the chart of the BGP table here
http://bgp.potaroo.net/ seems more like a power
curve than the exponential curve prior to 2001.
--Michael Dillon
--Michael Dillon
More on BGP table size and the number of fragmentary
announcements in the Internet
http://www.tm.uka.de/idrws/2004/contributions2004/IDRWS2004--04--Huston_Geoff--Allocations_and_Advertisements.pdf
This is Geoff Huston's presentation at the Inter-Domain
Routing Workshop in May 2004. Slides for all the
other talks here http://www.tm.uka.de/idrws/2004/contributions2004/
I suggest that if people find this stuff useful, they
might want to ask the authors to provide an update
at the next NANOG meeting.
As for Joe's lament about no new studies, I think
that after it was pointed out that BGP table growth
had halted http://www.netsys.com/library/papers/cengiz-bgp-2002-08.pdf
many people probably thought that the problem had
been solved forever by the telecom collapse.
--Michael Dillon
The author is providing a service by giving us raw data. If that is all they want to do, we cannot (and should not) force them to do more. Besides, I like raw data. 
Also, as for the "knee" Joe mentioned, I think he is talking about the fact the report went wonky. Look at the data presented in the last CIDR report - it is nonsense, obviously in error. This is not the "shape" of the "curve", it is the data itself.
[my attribution clipped -jzp]
>>- this month, another knee was at 150k [Dec 4th] and similarly
>> garbled results came out. Again, no response.
>>...in this one year we've seen the shape of the climb return to the
>>curve characterized by two years 99-01. Going for e? I'm not quite
>>sure what the current point of the report is if no-one responds to
>>even it breaking.
>
>Knee? Shape? Curve? Are you reading the same CIDR report
>that I see here every Friday? The report that I see is
>basically a dump of raw data. Perhaps the author needs
>to remember the distinction between data and information
>and make the CIDR report into something that people
>*WANT* to read. This posting of yours contained far more
>information than any CIDR report.
[snip]
Also, as for the "knee" Joe mentioned, I think he is talking about the
fact the report went wonky. Look at the data presented in the last
CIDR report - it is nonsense, obviously in error. This is not the
"shape" of the "curve", it is the data itself.
Correct on 'knee' but for crying out loud, follow the pointy clicky
references to the website. Of course there isn't going to be a curve
in email [you want ascii plots? how 1980s], but the email quite
clearly points you the way to the site where there is some analysis
of the raw data.
For many of us, the mail is a reminder 'here's the current raw info,
check the detailed stuff over here'. There is no secret cabal or
hidden ionfo, the report email
Joe, finding it a sad state of affairs that he must cut and paste
"http://www.cidr-report.org/cgi-bin/plota?file=%2Fvar%2Fdata%2Fbgp%2Fas4637%2Fbgp-active.txt&descr=Active%20BGP%20entries%20(FIB)&ylabel=Active%20BGP%20entries%20(FIB)&with=step" into this message for people to
actually look at the graph.
PS "2001 bellovin bush griffin rexford" entered into google hits the
specific reference quite nicely - sorry i didn't include the title.
as michael pointed out the specific links migrate all the time, so
i was purposefully avoiding 'where it can be found at this moment'
[try http://www.research.att.com/~jrex/papers/filter.pdf]
Correct on 'knee' but for crying out loud, follow the pointy clicky
references to the website. Of course there isn't going to be a curve
in email [you want ascii plots? how 1980s], but the email quite
clearly points you the way to the site where there is some analysis
of the raw data.
My bad 
But I include the CIDR reports website in my
complaint about data versus information. Yes
it does have SOME analysis and that is good.
But the way it is presented overwhelms one with
data and obscures the point of the website.
Also, I somehow missed the URL for the plot
that you posted even though I've been to this
website several times.
In fact, it shows that when Cengiz/Packeteer
presented the findings showing almost no growth,
the routing table was about to begin growing
again at the same rate as prior to the telecom
collapse. Packeteer's data did get a certain
amount of press coverage, Lightreading for instance,
so maybe that's why most people stopped looking
at how to control routing table growth.
However, CAIDA did make a presentation about
atoms in the AS path last fall
http://www.caida.org/projects/routing/atoms/documents/atoms-widew0311.pdf
If only everyone ran their BGP processes
on servers rather than routers, people could
actually begin using this now because the code
is available http://www.caida.org/projects/routing/atoms/
Now that, whether you agree with the approach
or not, is definitely something new in regards
to managing routing table growth.
--Michael Dillon