tlug: Will the REAL Japanese Internet please connect to Tsukuba?

["Stephen J. Turnbull" <turnbull@example.com>]

--------------------------------------------------------
tlug note from "Stephen J. Turnbull" <turnbull@example.com>
--------------------------------------------------------
>>>>> "Craig" == Craig Oda <craig@example.com> writes:

    Craig> On Mon, 12 May 1997, Steve Turnbull wrote:

    >> :-( "Japanese Internet" :-P is an oxymoron.  )-:

    Craig> The Japanese Internet is getting faster.  Systems connected
    Craig> to NSPIXP1 and especially NSPIXP2 are really getting fast.

Individual backbones are getting fast.  "Internet" means crossing
backbones.

    Craig> I'm getting 40 ms ping times between TWICS and your ISP
    Craig> through NSPIXP2.  I think that you'll see fast connection
    Craig> between Universities in Japan on Sinet.  So, I imagine that

Sure, inside of SINET connections are fast.

    Craig> Stephen can suck down data from Tokyo University pretty
    Craig> fast.  However, the gateway between the Japanese academic
    Craig> and the rest of the Internet is slow.

It's worse than that, unfortunately.  As far as I can tell, all of the
main backbone providers in Japan fumble whenever they need to hand off
to another one.  Certainly interconnections between WIDE and Sinet and
NSPIXP and Sinet are horrible; a year ago I was having pretty bad luck
with crossing the NSPIXP to WIDE boundary; I haven't tried that
recently.  It is not unusual for me to do a traceroute to gol1.gol.com
and discover that ping times to Sinet's Otsuka gateway are maybe 12ms,
that the Tokyo WIDE gateway (next hop) is 80-120ms with one packet
dropped, get no pings from the NSPIXP gate (2 hops from Sinet-Otsuka),
and have 150-300ms pings inside of GOL.

Contrast this with going to the US.  Dropped packets mostly occur on
the satellite link to Sprint's Stockton CA site; transfers to MCI and
BBN Planet do not require noticeably more time at the interconnect
point in San Jose.  (Average ping times to the next provider at
MaeWest are 2-5ms greater than Sprint's terminal at MaeWest.)

By the way, I often use GOL to get files from the US, since they have
a direct connection (with PSI, I think) to the US backbone.  The hard
part is shipping files to Tsukuba from Tokyo.  Early in the morning
it's pretty reliable, although still slow.

    Craig> Personally, I'm using ftp.lab.kdd.co.jp all the time and it
    Craig> feels like their server is on my ethernet.  :-) Well,
    Craig> almost...

KDD connects to Sinet through something called IMNET (Inter-Ministry
Research Informaion Network according to whois.nic.ad.jp).  This looks
like a North American MAE, with 5 media bridges inside of IMNET alone
on the route from Tsukuba.  (I'm assuming that sites labeled "FDDI"
and "ATM" and so on do connect via the media in their names....)  That
looks like one slick operation.  But they don't seem to be passing on
the benefits to the commercial / regular government users yet (pings
from Tsukuba to www.sorifu.go.jp and www.etl.go.jp both go through
WIDE, with 350-800ms times ;-)  I think it bodes well for the future,
though.

-- 
                            Stephen J. Turnbull
Institute of Policy and Planning Sciences                    Yaseppochi-Gumi
University of Tsukuba                      http://turnbull.sk.tsukuba.ac.jp/
Tel: +81 (298) 53-5091;  Fax: 55-3849              turnbull@example.com

-----------------------------------------------------------------
a word from the sponsor will appear below
-----------------------------------------------------------------
The TLUG mailing list is proudly sponsored by TWICS - Japan's First
Public-Access Internet System.  Now offering 20,000 yen/year flat
rate Internet access with no time charges.  Full line of corporate
Internet and intranet products are available.   info@example.com
Tel: 03-3351-5977   Fax: 03-3353-6096