Sentech's Crossed wires

regardtv · Aug 20, 2004

I've noticed that the routing on incomming traffic vs routing on outgoing traffic has undergone some changes af late - anyone else notice this ?

Data travelling TO MySmokesignal:

6 351 ms 410 ms 411 ms isl.nmszone.is.co.za [196.26.0.3]
7 371 ms 210 ms 210 ms 168.209.18.62
8 310 ms 201 ms 200 ms 66.18.87.251
9 380 ms 421 ms * 66.18.67.26
10 * 150 ms 191 ms 66.18.67.106
11 170 ms 180 ms 231 ms 66.18.87.50
12 741 ms 1032 ms 721 ms myw-stp-66-18-81-73 [66.18.81.73]

Data traveling FROM MySmokeSignal:

1 66.18.87.50 (66.18.87.50) 79.791 ms 183.077 ms 497.345 ms
2 66.18.67.106 (66.18.67.106) 179.964 ms 79.884 ms 122.663 ms
3 66.18.67.105 (66.18.67.105) 243.277 ms 239.453 ms 141.233 ms
4 66.18.67.25 (66.18.67.25) 140.598 ms * 265.574 ms
5 66.18.87.249 (66.18.87.249) 229.789 ms 199.043 ms 260.735 ms
6 66.18.87.241 (66.18.87.241) 242.100 ms 200.753 ms 239.325 ms
7 66.18.87.233 (66.18.87.233) 200.975 ms 121.579 ms *
8 66.18.87.225 (66.18.87.225) 178.995 ms 201.138 ms 120.460 ms
9 168.209.18.61 (168.209.18.61) 165.582 ms 364.720 ms 299.959 ms
10 196.26.0.253 (196.26.0.253) 139.262 ms * 240.324 ms

Notice the number of bounces from 1-->9 on the FROM section ...

Any ideas comments ?

<hr noshade size="1"><div align="right">8:00 Thursday 26 August 2004 is D-Day....
<s> 16 17 18 19</s> 20 21 20 23 24 25 26
</div id="right">

chuckl · Aug 20, 2004

Can't really answer fully rtv, but on tracert I'm 2 hops from 66.18.81.73 and 1 from 66.18.87.50.

www.sentech.co.za however, has gone from 6 to 7 hops with the 'tuning'

If we don't take care of the customers, maybe they'll stop bugging us.

regardtv · Aug 20, 2004

Can anyone with a 256K package run a trace route to www.mweb.co.za ... need to test a theory ...

<hr noshade size="1"><div align="right">8:00 Thursday 26 August 2004 is D-Day....
<s> 16 17 18 19</s> 20 21 20 23 24 25 26
</div id="right">

guest2013-1 · Aug 20, 2004

regard, I know what you're thinking... and I doubt it

To me 9 hops just to get OUT of the freaking network to get some data is unacceptable...

Sentech, 99% downtime guarunteed!

regardtv · Aug 21, 2004

There's something else comming noone .. trying to see how 256 routes vs how 128/512 routes....

<hr noshade size="1"><div align="right">8:00 Thursday 26 August 2004 is D-Day....
<s> 16 17 18 19 20</s> 21 20 23 24 25 26
</div id="right">

Nightwatch · Aug 21, 2004

On 256: Don't worry about 1, it's my own...

1 * * * Request timed out.
2 292 ms 140 ms 203 ms 66.18.87.50
3 105 ms 124 ms 109 ms 66.18.67.106
4 74 ms 124 ms 249 ms 66.18.67.105
5 167 ms 171 ms 249 ms 66.18.67.25
6 308 ms 171 ms 422 ms 66.18.87.249
7 73 ms 140 ms 187 ms 66.18.87.241
8 370 ms 281 ms 187 ms 66.18.87.233
9 230 ms 343 ms 406 ms 66.18.87.225
10 229 ms 202 ms 203 ms 168.209.18.61
11 136 ms 187 ms 187 ms 196.26.0.253
12 136 ms 156 ms 187 ms fe5-0.br1.jnb7.alter.net [196.30.228.241]
13 167 ms 171 ms 109 ms fe3-0.cr1.jnb7.alter.net [196.31.17.161]
14 167 ms 124 ms 281 ms fe0-0.gw2.cpt2.alter.net [196.30.205.4]
15 214 ms 156 ms 124 ms mweb-cpt2-1.cust-gw.cpt2.alter.net [196.31.167.110]
16 144 ms 200 ms 159 ms www.mweb.co.za [196.2.63.110]

<div align="right">DarkLigth Nocturnal Entertainment</div id="right">

anakin · Aug 21, 2004

Slighly offtopic. I had assumed that my 15%-20% packet loss was due to the air interface to my tower on 128k.

Code:

tracert www.sentech.co.za

Tracing route to www.sentech.co.za [66.18.65.115]
over a maximum of 30 hops:

  1     *        *        *     Request timed out.
  2   105 ms    99 ms    80 ms  66.18.87.50
  3    97 ms    99 ms   179 ms  66.18.67.106
  4    71 ms   148 ms    60 ms  66.18.67.105
  5   227 ms    89 ms    79 ms  66.18.67.25
  6    86 ms    99 ms    79 ms  66.18.87.250
  7    67 ms    79 ms    89 ms  pf-hosting1b-stp.infosat.net [66.18.67.203]
  8    76 ms   329 ms   101 ms  www.sentech.co.za [66.18.65.115]

However, until 5th. hop to 66.18.67.25, I experience no packet loss. Is this hop a microwave backhaul? Packet loss seems to get progressively worse from there.

<hr noshade size="1"><center>
My views are mine, and mine alone.</center>

dbnnet · Aug 21, 2004

The problem is not the air interface (tower).
Try doing tests after a major network failure
(or when Telkom looses it's International links again)
- There is much reduced core network activity and Packet loss goes
to almost zero.

Bay of Plenty : Signal 48% - SNL 17 - ber 71% : D-Link DI-604 - PPPoE : 8dBi Patch Antenna : Firmware 4.2.1.8

regardtv · Aug 21, 2004

Agreed there dbnnet ... I'm leaning towards hardware managing the bandwidth not able to sustain the load ...

<hr noshade size="1"><div align="right">8:00 Thursday 26 August 2004 is D-Day....
<s> 16 17 18 19 20</s> 21 20 23 24 25 26
</div id="right">

dbnnet · Aug 21, 2004

You've now put a picture in my head......
This 200 million rand IPW network
all linked through to a single Cisco 1601
router in JHB [}

]

Bay of Plenty : Signal 48% - SNL 17 - ber 71% : D-Link DI-604 - PPPoE : 8dBi Patch Antenna : Firmware 4.2.1.8

guest2013-1 · Aug 21, 2004

with the corporate monkeys, it won't be such a shock to me if they did... bwhahaha

Sentech, 99% downtime guarunteed!

dorris · Aug 21, 2004

Last time I spoke to the call centre about my packetloss ticket, It dawned on the ogre to ask me, what do you think the cause of this is??
I suggested a) the core router isn't coping with the load
b) microwave links can't cope with tower demands.

I noticed a pattern up till a week ago, where, latency would be quite high, I would then see 1/2 lost packets, following which latency would drop to sub 100.
Gave me the feeling that the congested item was just resetting itself , and resuming after on a clean slate.
this seems to have changed recently, now theres no pattern, its just erraticly BAD!!!

Interesting test I tried yesterday, pinged a mywireless account from ADSL, avg rtt was around 180
today, pinged the same ADSL account from mywireless, avg rtt 721

Another big fat WTF

Although, if all the routing tables were changed between yesterday and today, it could account for the change in latency from 2 ends of the connection, perhaps an adsl user can try ping a mywireless user.
If anyone wants to attempt my connection, my firewall will be responding to ICMP's until tomorrow morning MW 128, dorris.no-ip.com

dorris · Aug 22, 2004

Things have reverted to what they should be, going for the same adsl ip address again this morning, avg times were 170ms, which is what I would have expected.
so something changed between last night and this morning. PacketLoss has also dropped substantially.
probably just sunday morning, people still too hungover to be downloading pr0n.

Just a thought on the new routes chosen, perhaps it is some new mechanism in order to balance the load, over the congested areas, creating backup routes for the times when things seem too congested.
On 2nd thought, this sounds too complex for sentech.

guest2013-1 · Aug 22, 2004

i start downloading p0rn BEFORE drinking.... pffff, amateurs

Sentech, 99% downtime guarunteed!

Sentech's Crossed wires

regardtv

Expert Member

chuckl

Well-Known Member

regardtv

Expert Member

guest2013-1

guest

regardtv

Expert Member

Nightwatch

Member

anakin

Well-Known Member

dbnnet

Senior Member

regardtv

Expert Member

dbnnet

Senior Member

guest2013-1

guest

dorris

Well-Known Member

dorris

Well-Known Member

guest2013-1

guest