chip@vector.UUCP (Chip Rosenthal) (06/25/88)
On June 10, I broadcasted a "sendsys" message through the pubnet. I have
accumulated all of the responses. In summary, I have found:
- There are 96 systems receiving the pubnet distribution.
- There are two pubnet backbones.
To determine the pubnet connectivity, I hacked together some scripts to
generate statistics from the sys files. From this information, I manually
created a "backbone" map. In a moment, I'll describe my definition of a
backbone site, but first, here is the map:
bu-cs ----- UUNET ----- lll-winken ----- DDSW1 ----- ditka
petro ----- tness7 ----- KILLER ----- bigtex ----- qiclab
I defined a "backbone" site as a site which exchanges all messages within
at least one pubnet newsgroup (e.g. a link with the 'L' sys file flag is
not counted) with more than five other sites, and at least one of those
other sites is a "backbone" site. A "major backbone" site exchanges
pubnet messages with at least ten other sites. Major backbones sites are
capitalized in the map.
By this definition, there are two backbones to pubnet. I went through the
paths of all my pubnet news articles to see how messages are crossing
from the UUNET/DDSW1 backbone to the KILLER backbone. It appears that
messages are going DDSW1-->KILLER via two intermediate sites:
ddsw1!lll_winken!netsys!killer.
It appears that messages are going UUNET-->KILLER by four intermediate
sites (ick):
uunet!warble!rpp386!dalsqnt!pollux!killer.
By the way my site is on the KILLER backbone, which is why I'm focusing
on the -->KILLER flow.
I don't think this connectivity is too great. There are two things
which would fix this. First, it would be nice if some site could connect
between killer and uunet. Second, we should promote netsys to an honorary
backbone site :-).
Here are all the sites which responded to the sendsys request:
ausG.UUCP fed.FRB.GOV mcdchg.UUCP spike.UUCP
bacchus fluke.COM mcsd.UUCP spl1.UUCP
beartrk.UUCP glakes.UUCP mjbtn.UUCP spt.entity.com
bellboy.UUCP grebyn.COM mrsvr.UUCP stb.UUCP
bigtex.uucp helps.UUCP murphy.UUCP steinmetz.ge.com
bloom-beacon.MIT.ED hi.unm.edu mybest.UUCP studsys.mu.edu
bu-cs.BU.EDU icus.UUCP ncoast.UUCP sugar.UUCP
bucsb.UUCP igloo.UUCP ncsuvx.ncsu.edu sw1e.UUCP
buengc.BU.EDU iitmax.IIT.EDU netsys.UUCP swlabs.UUCP
bywater.UUCP illian.UUCP ninja.UUCP ti-csl.CSNET
catlabs.UUCP inco.UUCP nosun.UUCP tis.llnl.gov
chinet.UUCP inta1.UUCP obdient.UUCP tness1.UUCP
compsys.mu.edu iquery.UUCP otishq.UUCP tness7.UUCP
crash.cts.com irs3.UUCP pacbell.PacBell.COM ubvax.UB.Com
crlabs.UUCP jackson.UUCP pcmsam.UUCP unisec.usi.com
daitc.ARPA jclyde.UUCP percival.UUCP usna.MIL
dalsqnt.UUCP jolnet.UUCP petro.UUCP uunet.UU.NET
dasys1.UUCP juniper.UUCP pollux.UUCP uwmcsd1.UUCP
ddsw1.UUCP killer.UUCP portal.UUcp vdsvax.steinmetz.ge
ditka.UUCP lakesys.UUCP psu-cs.UUCP vector.UUCP
eddie.MIT.EDU ll1a.UUCP qiclab.UUCP vsi1.UUCP
eklektik.UUCP lll-lcc.aRpA richp1.UUCP warble.UUCP
elroy.Jpl.Nasa.Gov lll-winken.llnl.gov rpp386.UUCP wybbs.UUCP
etnibsd.UUCP marque.mu.edu sausys.UUCP zeek.UUCP
Following is my data on the sites which feed pubnet to at least five other
sites. Remember, links via the 'L' flag are not counted. Also, my scripts
aren't intelligent enough to discern a real site from a phony sys file
entry (such as an entry for a pubnet.sources archiver).
killer 17 backup,ames,bigtex,chinet,rpp386,dcs,icus,jolnet,mcsd,
mit-eddie,mjbtn,netsys,ozdaltx,tness7,wybbs,vector,test-bounce
uunet 15 portal,usna,bywater,fed,steinmetz,hnsurg3,altnet,warble,wash08,
etnibsd,cwi,pi,inco,island,SOURCES
ddsw1 11 atrium,chinet,ditka,gryphon,igloo,indep1,lll-winken,obdient,
spl1,swlabs,tapa
qiclab 10 psu-cs,percival,larkint,catlabs,bigtex,pcmsam,ohsu-hcx,mntgfx,
illian,nosun
bigtex 10 juniper,jclyde,helps,mybest,natinst,unisec,chinet,killer,
qiclab,mybest
tness7 9 tness1,swbatl,ness386,ninja,texas1,killer,petro,bellcore,bellboy
lll-winken 9 uunet,ubvax,scooter,ncdpc1,gryphon,netsys,hc.dspo.gov,lll-tis,spl1
marque 7 compsys,studsys,uwmcsd1,lakesys,madnix,gryphon,introl
petro 6 swri14be,iquery,swrinde,tness7,sausys,master
ditka 6 dasys1,bacchus,ddsw1,mcdchg,eklektik,stb
bu-cs 6 uunet,ncsuvx,bloom-beacon,cloud9,buengc,bucsb
tis 5 lll-lcc,lll-ncis,lll-winken,mordor,pacbell
ninja 5 tness7,pollux,killer,dalsqnt,sys1
--
Chip Rosenthal /// chip@vector.UUCP /// Dallas Semiconductor /// 214-450-0400
{uunet!warble,sun!texsun!rpp386,killer}!vector!chip
I won't sing for politicians. Ain't singing for Spuds. This note's for you.werner@utastro.UUCP (Werner Uhrig) (06/26/88)
Chip, that's a nice, informative report; there was only one thing I could not figure out: What is PUBNET? the article didn't say, and I don't recall the definition from earlier articles to new.admin. may be my aging brain-cells ... --Werner PS: when I did a "follow-up", posting failed as follows: What now? [send, edit, list, quit, write] s Posting article... inews: Unknown newsgroup pubnet.config Article not posted - exit status 256 I'll try again after changing the order of the groups in the Newsgroups-header from: Newsgroups: pubnet.config,news.admin to Newsgroups: news.admin,pubnet.config should that be classified as a BUG or a FEATURE ? :-)( <= not sure which ..... the bad news is, it didn't work either, so now I have to remove pubnet.config from the list ... apologies to the readers of that group (who also read news.admin; the heck with the others ..:-)