🚀 go-pugleaf

neil@blitzen.canberra.edu.au (Neil Symons (944199)) writes:
>CaT (cat@zip.com.au) wrote:
>> spakman@blitzen.canberra.edu.au (Ben L Trinder (954479)) writes:

>> >Hi, what is the deal with the new gateway(s). Here in canberra, we used
>> >to have a good link to the rest of the world (6Mbit to Melbourne). Now it
>> >seems all our international links go from Canberra, via Sydney to
>> >Melborne at 2 Mbit...  Why would anyone upgrade to a slower system?

>> >Also, since the new system, I have not been able to telnet nor rlogin to
>> >either ozemail.com.au, terminal.ozemail.com.au nor shell01.ozemail.com.au
>> >from any machine in canberra.edu.au. Is this a coincidence or is the new
>> >gateways really screwed up?

>> You could be trying to go through national-new.gw.au which, the last time I
>> had to try and use it, was fucked up in a big way for 4 days (and still may
>> be - I'm counting from Tuesday, 6-7pm).

>Well we still are and its still shocking.


>> Telstra suck and thank god my service provider ain't with em.

>Did you know that it took 30 hops with "traceroute" to get to a site in 
>the US near the Sydney->US gateway.
>When before via Melbourne it took less hops.

>I've forgotten the site so I will now do a traceroute to say a popular 
>bbs called isca bbs.  (whip.isca.uiowa.edu)

>traceroute to whip.isca.uiowa.edu (128.255.200.69), 30 hops max, 40 byte 
>packets
> 1  isegw.canberra.edu.au (137.92.11.50)  15.106 ms  2.18 ms  2.237 ms
> 2  AARnet.canberra.edu.au (137.92.1.90)  4.107 ms  5.12 ms  9.184 ms
> 3  act.gw.au (139.130.112.1)  986.032 ms  1175.09 ms  704.223 ms
> 4  national-new.gw.au (139.130.249.1)  1094.89 ms  1094.55 ms  788.757 ms
> 5  border1-serial3-5.LosAngeles.mci.net (204.70.171.61)  1271.77 ms  1212.07 ms  1136.32 ms
> 6  core1-fddi-0.LosAngeles.mci.net (204.70.170.17)  1157.95 ms  1018.06 ms  985.008 ms
> 7  core-hssi-2.SanFrancisco.mci.net (204.70.1.153)  1824.81 ms  616.582 ms  642.398 ms
> 8  * * core2-aip-4.SanFrancisco.mci.net (204.70.1.54)  669.553 ms
> 9  core1-hssi-2.Sacramento.mci.net (204.70.1.146)  645.76 ms  531.95 ms  534.395 ms
>10  core1-hssi-2.WillowSprings.mci.net (204.70.1.174)  732.84 ms * *
>11  core-hssi-3.Chicago.mci.net (204.70.1.85)  1122.02 ms * *
>12  border2-fddi-0.Chicago.mci.net (204.70.3.82)  1225.47 ms  1144.29 ms *
>13  cicnet.Chicago.mci.net (204.70.25.10)  762.41 ms *  970.991 ms
>14  dgf-fddi1.cic.net (192.217.0.4)  720.159 ms  1007.52 ms  1001.72 ms
>15  dgc-fddi0.cic.net (131.103.1.18)  957.891 ms *  845.17 ms
>16  * * uia-dgc.cic.net (131.103.26.34)  667.388 ms
>17  * * 204.42.128.130 (204.42.128.130)  1051.34 ms
>18  whip.isca.uiowa.edu (128.255.200.69)  820.826 ms  904.039 ms  895.403 ms

>I thought if you go to 16, then its not worth connecing to it all.

>Obviousely most of them are from US, but my point is that when we 
>(canberra uni) went via Melbourne, there were less hops. (and much faster)

Dont forget, all incoming traffic to TIS from the USA (and I
think Europe) is coming in via the old 6Meg link via melbourne.  Here
is a traceroute from the USA end...

peter@freefall.cdrom.com[1:48am]~-101> traceroute blitzen.canberra.edu.au
traceroute to blitzen.canberra.edu.au (137.92.11.168), 30 hops max, 40 byte packets
 1  gatekeeper.cdrom.com (192.216.222.1)  2.139 ms  1.954 ms  2.861 ms
 2  ucb2.barrnet.net (131.119.70.133)  3.130 ms  3.131 ms  3.108 ms
 3  ucbn1.barrnet.net (131.119.253.5)  4.110 ms  3.920 ms  11.277 ms
 4  su-b.barrnet.net (131.119.2.1)  5.315 ms  9.502 ms  6.248 ms
 5  cpe1-fddi-0.hay.mci.net (192.31.48.245)  6.501 ms  13.170 ms  5.593 ms
 6  borderx1-hssi2-0.SanFrancisco.mci.net (204.70.158.101)  7.837 ms  7.626 ms  12.713 ms
 7  core-fddi-0.SanFrancisco.mci.net (204.70.2.161)  20.588 ms  18.201 ms  24.134 ms
 8  border4-fddi-0.SanFrancisco.mci.net (204.70.3.163)  7.649 ms  7.532 ms  13.458 ms
 9  aarnet.SanFrancisco.mci.net (204.70.35.6)  196.438 ms  188.199 ms  191.951 ms
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10  * national-aix-us.gw.au (139.130.29.1)  207.283 ms  203.449 ms
    ^^^^^^^^^^^^^^^^^^^^^^^
11  act.gw.au (139.130.188.2)  466.119 ms  474.751 ms  466.319 ms
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
12  uc.gw.au (139.130.112.2)  769.006 ms  791.254 ms  802.608 ms
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
13  * * isegw.canberra.edu.au (137.92.1.85)  1072.760 ms
14  blitzen.canberra.edu.au (137.92.11.168)  1025.406 ms  889.412 ms  927.890 ms
peter@freefall.cdrom.com[1:49am]~-102>

Note that all traffic is going via San Francisco/Melbourne
(national-aix-us.gw.au is a name for the USA interface on national.gw.au)
and then up to canberra (where it gains another 250ms delay).

The other odd bit is that the delay going from the mci core border to
the aarnet router in SanFrancisco is incurring a large hit.. Could
that be flooded too perhaps?

The routing is asymmetric.. Even though your packets are going to the
USA via national-new.gw.au and Los Angeles, they are returning via the
slow congested link.

I suspect that TIS are going to have to bite the bullet and split the
australian bgp "area" into two parts...   Global routing table size
shouldn't be an issue, because A) it's contained entirely within MCI
and the routes would be aggregated back together as they left MCI,
and B) Access One have got their own link to MCI with selective IP
addresses from the 203.* netblk going to them - the precedent is
already there.. "It can be done"....

-Peter

>-- Neil
>-=-=-=-=-=-=-=-=-=-=-=-neil@blitzen.canberra.edu.au-=-=-=-=-=-=-=-=-=-=-=-
>Microsoft Network is prohibited from redistributing this work in any form,
>in whole or part, without a written license. Licence is available for
>$500 AUS. Transmission without a current licence constitutes that they 
>agree to these terms.
>-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-