95 replies to this topic

[Jon Cunningham]

A User's Guide to Interpretting Traceroutes to our systems.

Q. Here's a sample traceroute from a user. Router 4 times out! Is that okay?

Quote

1 1 ms 1 ms 1 ms dsldevice.lan [127.0.0.1]
2 20 ms 22 ms 22 ms lo0.10.central10.ptw-bng01.plus.net [195.166]
3 20 ms 14 ms 16 ms irb.10.PTW-CR01.plus.net [84.93.249.1]
4 * * * Request timed out.
5 19 ms 19 ms 17 ms gra-g2-a9.fr.eu [178.33.103.218]
6 18 ms 19 ms 17 ms 37.187.36.162
7 * * * Request timed out.
8 97 ms 96 ms 97 ms mwomercs.com [192.99.109.192]

A. Yes! Absolutely. Many core routers often don't respond to pings, they're too busy sending traffic. If you can see, the traffic from hop 3 to hop 5 gets there just fine.

-----------------------


Q. I used some advanced software and it shows that 60% of the traffic is being dropped!

Quote

A. That's also fine! Many routers simply don't prioritize responding to pings and drop most of them! If it was actually dropping 60% of your traffic, how would 100% of your traffic still make it to the next node?

---------------------------


Q. I pinged a random router in the traceroute and it didn't respond!

A. That's also fine! Why would it respond to a random ping? If every router on the internet had to respond to pings, folks would be able to do massive DDoS attacks on core components of the internet just by pinging them!

----------------------------

Q. I did a traceroute and one of the routers is taking way too long to respond!

Quote

1 1 ms 1 ms 1 ms dsldevice.lan [127.0.0.1]
2 20 ms 22 ms 22 ms lo0.10.central10.ptw-bng01.plus.net [195.166]
3 20 ms 14 ms 16 ms irb.10.PTW-CR01.plus.net [84.93.249.1]
4 * * * Request timed out.
5 900 ms 867 ms 757 ms gra-g2-a9.fr.eu [178.33.103.218]
6 18 ms 19 ms 17 ms 37.187.36.162
7 * * * Request timed out.
8 97 ms 96 ms 97 ms mwomercs.com [192.99.109.192]

A. Still fine. Same logic here - it's not prioritizing responding to pings because it has other more important stuff to do. Like making sure the traffic that passes through it gets where it is going. If you look, the time at hop 5 is really high, but if it really did take 900ms to pass any traffic, how is the traffic at hop 6 managing to come back in 18 ms?

---------------------------

Q. What would actually be a problem, then?

A. I'll give some examples from real problems

Quote

>tracert mwomercs.com
Tracing route to mwomercs.com [192.99.109.192]
over a maximum of 30 hops:
1 1 ms <1 ms <1 ms 192.168.1.1
2 30 ms 30 ms 29 ms b-ras1.lmk.limerick.eircom.net [159.134]
3 28 ms 28 ms 28 ms ge-1-1-7.pe1.lmk.lmk-mlw.eircom.net [86.43.247.33]
4 34 ms 39 ms 39 ms 86.43.253.72
5 34 ms 33 ms 33 ms tenge-2-2-1.core1.srl.core.eircom.net [86.43.252.54]
6 49 ms 46 ms 48 ms tenge-2-2-1.pe1.thn.the-thn.eircom.net [86.43.253.114]
7 49 ms 48 ms 45 ms ge7-1-0.corea.thn.london.eircom.net [86.43.244.194]
8 * * 60 ms sw1.tc.lon.ovh.net [195.66.224.220]
9 168 ms 170 ms 166 ms rbx-g2-a9.fr.eu [91.121.128.195]
10 177 ms 75 ms 73 ms 37.187.36.133
11 * * * Request timed out.
12 245 ms 249 ms 258 ms mwomercs.com [192.99.109.192]
Trace complete.

A1. This is a problem! Why is it taking a user in Ireland 177ms to reach our fibre in France when the average user in London can get there in 20ms?

Quote

1 <1 ms <1 ms <1 ms ntt.setup [192.168.1.1]
2 5 ms 6 ms 5 ms 122.28.24.38
3 7 ms 6 ms 6 ms 122.28.24.62
4 8 ms 8 ms 9 ms 221.184.12.221
5 6 ms 6 ms 6 ms 122.28.168.89
6 6 ms 6 ms 6 ms 118.23.168.89
7 13 ms 7 ms 7 ms ae-5.r21.tokyjp05.jp.bb.gin.ntt.net [129.250]
8 7 ms 9 ms 8 ms ae-0.r24.tokyjp05.jp.bb.gin.ntt.net [129.250.6.96]
9 81 ms 75 ms 81 ms ae-3.r21.sngpsi05.sg.bb.gin.ntt.net [129.250.7.35]
10 73 ms 73 ms 76 ms ae-13.r00.sngpsi02.sg.bb.gin.ntt.net [129.250.4.74]
11 75 ms 74 ms 73 ms ae0.singapore2.sin.seabone.net [93.186.133.12]
12 330 ms 333 ms 325 ms xe-2-0-3.milano51.mil.seabone.net [195.22.205.109]
13 329 ms * * mil-1-6k.it.eu [91.121.131.25]
14 * 333 ms * fra-1-6k.de.eu [213.251.128.66]
15 339 ms 337 ms 338 ms sbg-g1-a9.fr.eu [91.121.128.80]
16 * 340 ms 343 ms 37.187.36.153
17 * * * 要求がタイムアウトしました。
18 333 ms 322 ms 329 ms mwomercs.com [192.99.109.192]

A2. This is a problem! Why is this user in JAPAN being sent to Milano, Italy.

Let me know if you see issues like this. : )

There's some other types of problems that are a bit more complicated to discuss that I've left off the list here, but these are the common ones.

[CHH Badkarma]

Great to see useful, factual information.

Just an idea, maybe post up a "how to" on running a traceroute for those that don't know how.

[One of Little Harmony]

Where's the thread we post tracerts when they start looking funny?

[Telmasa]

Yay, noob-friendly technical information!

I'll just point out, the biggest indicator something is wrong seems to be the final hop to mwomercs. If that number is excessively high, something's going wonky - so then you know to look for earlier hops with suspiciously high pings.

If that number remains at a low/stable level, then things are pretty much fine.

[darkezero]

Figured I'd give it a shot

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| homeportal - 0 | 129 | 129 | 0 | 0 | 2 | 1 |
|108-251-136-2.lightspeed.frokca.sbcglobal.net - 43 | 47 | 27 | 19 | 20 | 23 | 20 |
| 75.29.64.134 - 0 | 129 | 129 | 20 | 21 | 27 | 21 |
| 12.83.77.137 - 0 | 129 | 129 | 20 | 23 | 54 | 24 |
| gar23.sffca.ip.att.net - 0 | 129 | 129 | 24 | 26 | 54 | 26 |
| No response from host - 100 | 25 | 0 | 0 | 0 | 0 | 0 |
| No response from host - 100 | 25 | 0 | 0 | 0 | 0 | 0 |
| 192.99.146.65 - 72 | 32 | 9 | 0 | 105 | 142 | 142 |
| mtl-1-6k.qc.ca - 26 | 63 | 47 | 0 | 118 | 463 | 100 |
| bhs-g1-6k.qc.ca - 0 | 129 | 129 | 96 | 100 | 127 | 105 |
| 198.27.73.231 - 0 | 129 | 129 | 96 | 100 | 132 | 101 |
| mwomercs.com - 0 | 129 | 129 | 97 | 101 | 134 | 105 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider


that mtl-1-6k.qc.ca is the current highest every time it eventually replies

[Surn]

Are we considering a second datacenter?

[White Bear 84]

idea: Information on how users who are non technical can identify where each IP address is located.

[Rushin Roulette]

CHH Badkarma, on 10 April 2015 - 12:14 PM, said:

Just an idea, maybe post up a "how to" on running a traceroute for those that don't know how.

The ususal way is to scream into your Microphone over VOIP with the standard Enquiry message "Bloody lag. why is PGI doing this to me?" and count the miliseconds until your teammates scream back with the standard Enquiry responce "Shut up, we are trying to concentrate here!"

The alternate way of doing this (is a little bit more complicated... but not much more) can be read here;
http://www.mediacoll...traceroute.html

Edited by Rushin Roulette, 14 April 2015 - 08:02 AM.

[Lily from animove]

MechregSurn, on 13 April 2015 - 07:10 PM, said:

Are we considering a second datacenter?

what for? even if you live next to it, when routing decides you go the other way around the world, it will hardly help

[Captain Fancy Pants]

Figured I'd show my results here:

C:\>tracert mwomercs.com

Tracing route to mwomercs.com [192.99.109.192]
over a maximum of 30 hops:

1 19 ms 7 ms 14 ms 192.168.0.1
2 10 ms 15 ms 8 ms 79.97.254.1
3 10 ms 9 ms 23 ms 79.97.254.1
4 37 ms 36 ms 24 ms 109.255.254.29
5 32 ms 33 ms 29 ms 84.116.238.70
6 32 ms 30 ms 30 ms nl-ams04a-ri2-xe-8-2-0.aorta.net [84.116.134.77]

7 31 ms 34 ms 32 ms nl-ams02a-rd2-xe-5-0-1.aorta.net [84.116.130.33]

8 31 ms 34 ms 28 ms 84.116.135.186
9 37 ms * 41 ms ams-5-6k.nl.eu [91.121.131.85]
10 41 ms 40 ms 37 ms rbx-g2-a9.fr.eu [178.33.100.230]
11 42 ms 41 ms 39 ms 37.187.36.133
12 * * * Request timed out.
13 134 ms 134 ms 135 ms mwomercs.com [192.99.109.192]

Trace complete.

[Morgana]

I can somewhat relate to the last example, although not to the extreme distance as it shows. But I'm scratching my head wondering why, from Maine, my route sends me clear down to New Jersey before it swings up toward Montreal. Considering travel time there and back, my packets are going 1000+/1 miles out of the way to get there. Seems though there should be a shorter way.

[Mike Forst]

Morgana, on 14 April 2015 - 10:56 AM, said:

I can somewhat relate to the last example, although not to the extreme distance as it shows. But I'm scratching my head wondering why, from Maine, my route sends me clear down to New Jersey before it swings up toward Montreal. Considering travel time there and back, my packets are going 1000+/1 miles out of the way to get there. Seems though there should be a shorter way.

This is all the "magic" of ISP networking. Odds are very good that the only available peering point your ISP has for Montreal is out of New Jersey. It's very unlikely that there are many if any peering locations in Maine. They probablt run most of it our of that one peering facility. It's really expensive to run fiber and most time it's just better (and certainly cheaper) to run it back through your network until it reaches the network egress.

The other thing to remember is that this is all traveling at the speed of light more or less so (using some quick math) all that extra distance is probably costing you about .4ms on your ping. That's not counting the extra latency from router hops and other stuff but overall the effect is probably fairly negligible.

[Angrah]

What I don't understand is the Southern cross undersea cable goes from Sydney to Nedonna Beach Oregon via Hawaii AND the Telstra Endeavour undersea cable goes to Hawaii as well.
Yet i'm being sent to Florida which has no direct cable connection to Australia.


Tracing route to mwomercs.com [192.99.109.192]
over a maximum of 30 hops:

1 1 ms <1 ms 1 ms home.gateway.home.gateway [192.168.1.254]
2 38 ms 52 ms 35 ms 172.18.208.9
3 35 ms 35 ms 36 ms 172.18.66.197
4 35 ms 36 ms 36 ms Bundle-Ether11.chw-edge901.sydney.telstra.net [1
39.130.190.165]
5 36 ms 36 ms 36 ms bundle-ether13.chw-core10.sydney.telstra.net [20
3.50.11.98]
6 37 ms 38 ms 39 ms Bundle-ether17.oxf-gw2.sydney.telstra.net [203.5
0.13.70]
7 39 ms 38 ms 39 ms bundle-ether1.sydo-core01.sydney.reach.com [203.
50.13.38]
8 38 ms 38 ms 38 ms 202.84.220.189
9 179 ms 178 ms 179 ms 202.84.249.50
10 179 ms 179 ms 178 ms 202.84.251.174
11 179 ms 178 ms 179 ms tisparkle-peer.eqla01.pr.telstraglobal.net [134.
159.61.26]
12 257 ms 256 ms 256 ms xe-1-3-0.miami15.mia.seabone.net [195.22.199.111
]
13 248 ms 256 ms 254 ms xe-0-0-0.miami15.mia.seabone.net [195.22.199.103
]
14 291 ms 291 ms * mia-1-6k.fl.us [178.32.135.208]
15 * * * Request timed out.
16 318 ms 317 ms 320 ms bhs-g1-6k.qc.ca [198.27.73.205]
17 301 ms 301 ms 334 ms 198.27.73.231
18 301 ms 300 ms 301 ms mwomercs.com [192.99.109.192]

Trace complete.

Edit: I have done some ip geolocating and it seems instead of going Sydney-Hawaii-USA(Oregon)-Canada it is going Sydney-Japan-Hong Kong-Italy-France-USA(Florida)-Canada

Edited by Angrah, 15 April 2015 - 01:46 AM.

[White Bear 84]

Angrah, on 14 April 2015 - 08:33 PM, said:

6 37 ms 38 ms 39 ms Bundle-ether17.oxf-gw2.sydney.telstra.net [203.5
0.13.70]
7 39 ms 38 ms 39 ms bundle-ether1.sydo-core01.sydney.reach.com [203.
50.13.38]
8 38 ms 38 ms 38 ms 202.84.220.189
9 179 ms 178 ms 179 ms 202.84.249.50
10 179 ms 179 ms 178 ms 202.84.251.174
11 179 ms 178 ms 179 ms tisparkle-peer.eqla01.pr.telstraglobal.net [134.
159.61.26]

Also interesting to note the jump in ping from 8-9?! Might be considered normal, but that is a huge difference which ends up affecting gameplay?! And this would appear to be BEFORE the undersea cable..



Although potentially answered by a Q above, the difference is that the ping after 9 does not drop back down - so my question is, if the ping is suddenly allocated a lower priority resulting in the lower response time within Australia, does this identify a potential bottleneck that would be directly affecting Aus pings?

Edited by White Bear 84, 14 April 2015 - 10:34 PM.

[Spiritwalker_]

This is mine

Tracing route to www.mwomercs.com [192.99.109.192]
over a maximum of 30 hops:

1 1 ms <1 ms <1 ms home.gateway.home.gateway [192.168.1.254]
2 8 ms 8 ms 8 ms loop0.lns20.drw1.on.ii.net [150.101.64.16]
3 8 ms 8 ms 8 ms gi1-30.cr1.drw1.on.ii.net [150.101.64.49]
4 57 ms * 59 ms gi1-1-0.cor2.bne4.on.ii.net [150.101.64.14]
5 142 ms 83 ms 164 ms ae1.br1.syd4.on.ii.net [150.101.33.18]
6 219 ms 219 ms 218 ms te0-1-1-3.br2.lax1.on.ii.net [150.101.33.195]
7 218 ms 218 ms 218 ms any2ix.coresite.com [206.72.210.214]
8 * * * Request timed out.
9 289 ms 286 ms 288 ms bhs-g2-6k.qc.ca [198.27.73.207]
10 286 ms 292 ms 286 ms 198.27.73.233
11 316 ms 291 ms 285 ms mwomercs.com [192.99.109.192]

Trace complete.

Is a little weird why goes to Adelaide then back up to border of SA and NT and then out to (this is WTF territory) Kansas?!? (should go LA I would think) then up to Canada from there.

[Mike Forst]

Angrah, on 14 April 2015 - 08:33 PM, said:

What I don't understand is the Southern cross undersea cable goes from Sydney to Nedonna Beach Oregon via Hawaii AND the Telstra Endeavour undersea cable goes to Hawaii as well.
Yet i'm being sent to Florida which has no direct cable connection to Australia.


Tracing route to mwomercs.com [192.99.109.192]
over a maximum of 30 hops:

1 1 ms <1 ms 1 ms home.gateway.home.gateway [192.168.1.254]
2 38 ms 52 ms 35 ms 172.18.208.9
3 35 ms 35 ms 36 ms 172.18.66.197
4 35 ms 36 ms 36 ms Bundle-Ether11.chw-edge901.sydney.telstra.net [1
39.130.190.165]
5 36 ms 36 ms 36 ms bundle-ether13.chw-core10.sydney.telstra.net [20
3.50.11.98]
6 37 ms 38 ms 39 ms Bundle-ether17.oxf-gw2.sydney.telstra.net [203.5
0.13.70]
7 39 ms 38 ms 39 ms bundle-ether1.sydo-core01.sydney.reach.com [203.
50.13.38]
8 38 ms 38 ms 38 ms 202.84.220.189
9 179 ms 178 ms 179 ms 202.84.249.50
10 179 ms 179 ms 178 ms 202.84.251.174
11 179 ms 178 ms 179 ms tisparkle-peer.eqla01.pr.telstraglobal.net [134.
159.61.26]
12 257 ms 256 ms 256 ms xe-1-3-0.miami15.mia.seabone.net [195.22.199.111
]
13 248 ms 256 ms 254 ms xe-0-0-0.miami15.mia.seabone.net [195.22.199.103
]
14 291 ms 291 ms * mia-1-6k.fl.us [178.32.135.208]
15 * * * Request timed out.
16 318 ms 317 ms 320 ms bhs-g1-6k.qc.ca [198.27.73.205]
17 301 ms 301 ms 334 ms 198.27.73.231
18 301 ms 300 ms 301 ms mwomercs.com [192.99.109.192]

Trace complete.

Edit: I have done some ip geolocating and it seems instead of going Sydney-Hawaii-USA(Oregon)-Canada it is going Sydney-Japan-Hong Kong-Italy-France-USA(Florida)-Canada

This once again comes down to peering and also just general internet routing. Your ISP's "best route" goes through Europe for whatever reason instead of across the Pacific. It's suboptimal for sure and is unfortunately once of those things where you're at their mercy. We can and do request that ISPs make routing changes to fix suboptimal stuff like this but they really don't have to listen to us at all. They might listen to you as a customer though and will more likely listen to the complaints of a bunch of customers.

[Spiritwalker_]

Mike Forst, on 16 April 2015 - 09:18 AM, said:

This once again comes down to peering and also just general internet routing. Your ISP's "best route" goes through Europe for whatever reason instead of across the Pacific. It's suboptimal for sure and is unfortunately once of those things where you're at their mercy. We can and do request that ISPs make routing changes to fix suboptimal stuff like this but they really don't have to listen to us at all. They might listen to you as a customer though and will more likely listen to the complaints of a bunch of customers.

Think this is Telstra ISP I think

[Moriquendi86]

Here is mine.

1 <1 ms <1 ms <1 ms 192.168.1.1
2 * * * Upłynął limit czasu żądania.
3 7 ms 7 ms 5 ms 89-75-20-146.infra.chello.pl [89.75.20.146]
4 7 ms 7 ms 7 ms pl-waw04a-rd2-ae13-2114.aorta.net [84.116.252.57]
5 6 ms 7 ms 7 ms 84.116.138.90
6 * 53 ms * var-1-6k.pl.eu [213.251.130.69]
7 * * * Upłynął limit czasu żądania.
8 * 45 ms 36 ms sbg-g1-a9.fr.eu [91.121.128.80]
9 43 ms 32 ms 31 ms 37.187.36.153
10 * * * Upłynął limit czasu żądania.
11 124 ms 127 ms 122 ms mwomercs.com [192.99.109.192]

Looks OK but someone has to do something about that ocean.

[Telmasa]

Just adding in - not really cause I have a problem, compared to other folks here, just curious about a couple things:

Tracing route to mwomercs.com [192.99.109.192]
over a maximum of 30 hops:
1 <1 ms 1 ms <1 ms 192.168.1.1
2 9 ms 9 ms 9 ms L100.LSANCA-VFTTP-121.verizon-gni.net [173.58.73.1]
3 15 ms 13 ms 13 ms G0-14-3-2.LSANCA-LCR-22.verizon-gni.net [130.81.131.44]
4 * * * Request timed out.
5 13 ms 37 ms 11 ms 0.ae2.XL4.LAX15.ALTER.NET [140.222.227.21]
6 17 ms 13 ms 15 ms TenGigE0-5-4-0.GW4.LAX15.ALTER.NET [152.63.115.246]
7 16 ms 14 ms 12 ms teliasonera-gw.customer.alter.net [152.179.21.42]
8 27 ms 26 ms 25 ms palo-b1-link.telia.net [80.91.248.191]
9 94 ms 93 ms 102 ms 192.99.146.32
10 * * * Request timed out.
11 87 ms 88 ms 86 ms mtl-1-6k.qc.ca [198.27.73.183]
12 104 ms 93 ms 95 ms bhs-g1-6k.qc.ca [198.27.73.1]
13 123 ms 135 ms 121 ms 198.27.73.231
14 99 ms 96 ms 93 ms mwomercs.com [192.99.109.192]


So, what I've discovered:
Hops 8 & 9, iplocation.net tells me, is the hop from Palo Alto (south of San Francisco) CA, over and up to you boys in Quebec. Not bad for going across the continent, eh? I'm assuming that's expectable. (edit: going back, previous hops possibly are going between Virginia, New York, and Las Vegas first though...this is confusing! I'm guessing it's really just bumping me around the L.A. basin over to San Fran, and the ip info is just a bit mixed up)

My question:
Hops 11, 12, and 13, all appear to be the same router tossing me around its servers...so it's also adding, what, 30-35 ping? (though ping falls again when it hops to mwomercs.com - not sure how, but that's neat lol)
What's the explanation for what's happening there? I'm scratching my head over it...especially cause my ping in-game appears as 121-135 more often than 93-96.

Edited by Telmasa, 20 April 2015 - 11:38 PM.

[Mike Forst]

spiritwalken, on 17 April 2015 - 11:14 PM, said:

Think this is Telstra ISP I think

Yes that does appear to be the case.