[Snap] Routing issues within Australia

Forums › 2degrees (including Slingshot, Orcon, Flip, Stuff Fibre, MyRepublic, 2talk and Vocus) › [Snap] Routing issues within Australia

Napster

202 posts

Master Geek

#100427 11-Apr-2012 14:56

Hi Guys,

I am having routing issues within Australia.
This is to some BF3 Servers hosted in Sydney.

Looks like its routing via Melbourne then back to Sydney.
Any of you having the same issues on Snap?

Check this out

kyhwana2

2566 posts

Uber Geek

#607875 11-Apr-2012 15:01

Looks fine to me?
If you're going to sydnet to melbourne and then back to sydney, yeah, you'll get some added latency. Looks like a telstra routing issue. Nothing you can do, really. (Or SNAP for that matter, all they can do is file a ticket with telstras NOC if there's actually an issue)

McGee

200 posts

Master Geek

#607883 11-Apr-2012 15:05

kyhwana2: Looks fine to me?
If you're going to sydnet to melbourne and then back to sydney, yeah, you'll get some added latency. Looks like a telstra routing issue. Nothing you can do, really. (Or SNAP for that matter, all they can do is file a ticket with telstras NOC if there's actually an issue)

rscole86

4973 posts

Uber Geek

Moderator

Trusted

Lifetime subscriber

#607885 11-Apr-2012 15:08

Not sure when I will be able to test this from the MIL snap connection, but for a comparison, here is my TNZ.

Tracing route to 173.199.80.80.choopa.net [173.199.80.80]
over a maximum of 30 hops:

1 1 ms 1 ms 1 ms homeportal [192.168.1.254]
2 35 ms 36 ms 34 ms 219-89-58-1.dialup.xtra.co.nz [219.89.58.1]
3 53 ms 47 ms 48 ms mdr-fid-int.akbr5.global-gateway.net.nz [202.37.244.222]
4 44 ms 44 ms 46 ms ae4-10.akbr5.global-gateway.net.nz [202.37.244.221]
5 44 ms 45 ms 44 ms xe3-1-0.tkbr9.global-gateway.net.nz [202.50.232.74]
6 67 ms 70 ms 70 ms xe0-1-0.sebr1.global-gateway.net.nz [202.50.232.14]
7 201 ms 226 ms 249 ms ae1-10.sebr2.global-gateway.net.nz [202.50.232.242]
8 70 ms 68 ms 69 ms internode.sebr2.global-gateway.net.nz [202.50.232.142]
9 70 ms 70 ms 69 ms te0-0-0.bdr1.syd7.internode.on.net [150.101.197.5]
10 70 ms 70 ms 70 ms gi1-2-104.cor2.syd7.internode.on.net [150.101.197.114]
11 69 ms 70 ms 70 ms gi0-0-1.pe2.syd7.internode.on.net [150.101.197.218]
12 70 ms 72 ms 69 ms eth59-167-137-1.static.internode.on.net [59.167.137.1]
13 70 ms 70 ms 70 ms 173.199.80.80.choopa.net [173.199.80.80]

Tracing route to 173.199.80.104.choopa.net [173.199.80.104]
over a maximum of 30 hops:

1 1 ms 1 ms 1 ms homeportal [192.168.1.254]
2 35 ms 34 ms 36 ms 219-89-58-1.dialup.xtra.co.nz [219.89.58.1]
3 * 1151 ms 1398 ms mdr-fid-int.akbr5.global-gateway.net.nz [202.37.244.222]
4 46 ms 46 ms 46 ms ae4-10.akbr5.global-gateway.net.nz [202.37.244.221]
5 43 ms 44 ms 48 ms xe3-1-0.tkbr9.global-gateway.net.nz [202.50.232.74]
6 69 ms 70 ms 68 ms xe0-1-0.sebr1.global-gateway.net.nz [202.50.232.14]
7 298 ms 298 ms 299 ms ae1-10.sebr2.global-gateway.net.nz [202.50.232.242]
8 69 ms 68 ms 70 ms internode.sebr2.global-gateway.net.nz [202.50.232.142]
9 69 ms 70 ms 70 ms te0-0-0.bdr1.syd7.internode.on.net [150.101.197.5]
10 69 ms 70 ms 70 ms gi1-2-104.cor2.syd7.internode.on.net [150.101.197.114]
11 70 ms 70 ms 70 ms gi0-0-1.pe2.syd7.internode.on.net [150.101.197.218]
12 69 ms 70 ms 72 ms eth59-167-137-1.static.internode.on.net [59.167.137.1]
13 70 ms 71 ms 70 ms 173.199.80.104.choopa.net [173.199.80.104]

frizianz

105 posts

Master Geek

#607890 11-Apr-2012 15:12

McGee:
kyhwana2: Looks fine to me?
If you're going to sydnet to melbourne and then back to sydney, yeah, you'll get some added latency. Looks like a telstra routing issue. Nothing you can do, really. (Or SNAP for that matter, all they can do is file a ticket with telstras NOC if there's actually an issue)

+1

DoomlordVekk

129 posts

Master Geek

Trusted

#608093 12-Apr-2012 01:07

Gents, with the risk of coming across as someone unsympathetic to your perceived problem, what I think you're seeing here isn't a routing problem, as maybe some sub-optimal peering arrangements. There certainly aren't any loops or black holes

Lets follow a packet...
Snap to TCL, via REACH/TI to Telstra's gateway router, all good there.
From there to one of the Telstra Kensington(?) core routers and then across to Chatswood core.

From here, it seems that there is a decision to head toward an Optus peering/Handover link in Melbourne is made.
seems a bit strange that Telstra doesn't peer with Optus in Sydney but that might be a business decision.

So, down to Melbourne, jump across to Optus who then hands it to Internode locally.
From there, Internode moves the traffic back to Sydney, where the endpoint network is located.

So 1)Why doesn't Telstra peer with Internode in Sydney?
2)Why does Telstra send traffic for Optus to Melbourne?

What you have to remember is you can only influence IP routing with BGP via what you can tell your next hop provider. They can then choose to A) Honour what you've asked them to do via various BGP attributes, B)send reacability info about your AS and prefixes to their upstreams as they want to see things or how they wish to control traffic to and from their upstreams into their networks.

SNAP can't control how Telstra tell Optus about how to move traffic for AS23655. All Optus knows is AS23655 is reachable via Telstra and Telstra have asked them to send traffic destined for AS23655 via this interconnect point or another one(Yes, it's over simplifying BGP, guilty as charged).
Snap could peer with Optus directly at Auckland but thats a $$ game for SNAP then.

IBM tried to make the world source route everything but it never took off and would never have scaled to become the Internet we know and get cranky with today.

"Customers don’t expect you to be perfect. They do expect you to fix things when they go wrong." Donald Porter – British Airways

The views expressed here are my own and are not reflective of other organisms or organisations.

mercutio

1392 posts

Uber Geek

#609284 14-Apr-2012 15:19

DoomlordVekk: Gents, with the risk of coming across as someone unsympathetic to your perceived problem, what I think you're seeing here isn't a routing problem, as maybe some sub-optimal peering arrangements. There certainly aren't any loops or black holes

Lets follow a packet...
Snap to TCL, via REACH/TI to Telstra's gateway router, all good there.
From there to one of the Telstra Kensington(?) core routers and then across to Chatswood core.

From here, it seems that there is a decision to head toward an Optus peering/Handover link in Melbourne is made.
seems a bit strange that Telstra doesn't peer with Optus in Sydney but that might be a business decision.

So, down to Melbourne, jump across to Optus who then hands it to Internode locally.
From there, Internode moves the traffic back to Sydney, where the endpoint network is located.

So 1)Why doesn't Telstra peer with Internode in Sydney?
2)Why does Telstra send traffic for Optus to Melbourne?

What you have to remember is you can only influence IP routing with BGP via what you can tell your next hop provider. They can then choose to A) Honour what you've asked them to do via various BGP attributes, B)send reacability info about your AS and prefixes to their upstreams as they want to see things or how they wish to control traffic to and from their upstreams into their networks.

SNAP can't control how Telstra tell Optus about how to move traffic for AS23655. All Optus knows is AS23655 is reachable via Telstra and Telstra have asked them to send traffic destined for AS23655 via this interconnect point or another one(Yes, it's over simplifying BGP, guilty as charged).
Snap could peer with Optus directly at Auckland but thats a $$ game for SNAP then.

IBM tried to make the world source route everything but it never took off and would never have scaled to become the Internet we know and get cranky with today.

well one solution would be for snap to peer at equinix.

telstra and telecom are renowned for having inadequete peering.

McGee

200 posts

Master Geek

#609285 14-Apr-2012 15:21

mercutio:

well one solution would be for snap to peer at equinix.

telstra and telecom are renown for having inadequete peering.

If only it was that simple eh.

Sharesies kids

Free kids accounts - trade shares and funds (NZ, US) with Sharesies (affiliate link).

mercutio

1392 posts

Uber Geek

#609287 14-Apr-2012 15:22

McGee:
mercutio:

well one solution would be for snap to peer at equinix.

telstra and telecom are renown for having inadequete peering.

If only it was that simple eh.

well it's always about money. but slingshot, and orcon are both doing it. maybe they can buy some capacity off one of them? even if they just sent their outgoing that way it'd reduce the latency. and they may be able to buy cheaper or less bandwidth for outbound only.

mercutio

1392 posts

Uber Geek

#609291 14-Apr-2012 15:30

DoomlordVekk: Gents, with the risk of coming across as someone unsympathetic to your perceived problem, what I think you're seeing here isn't a routing problem, as maybe some sub-optimal peering arrangements. There certainly aren't any loops or black holes

Lets follow a packet...
Snap to TCL, via REACH/TI to Telstra's gateway router, all good there.
From there to one of the Telstra Kensington(?) core routers and then across to Chatswood core.

From here, it seems that there is a decision to head toward an Optus peering/Handover link in Melbourne is made.
seems a bit strange that Telstra doesn't peer with Optus in Sydney but that might be a business decision.

So, down to Melbourne, jump across to Optus who then hands it to Internode locally.
From there, Internode moves the traffic back to Sydney, where the endpoint network is located.

So 1)Why doesn't Telstra peer with Internode in Sydney?
2)Why does Telstra send traffic for Optus to Melbourne?

What you have to remember is you can only influence IP routing with BGP via what you can tell your next hop provider. They can then choose to A) Honour what you've asked them to do via various BGP attributes, B)send reacability info about your AS and prefixes to their upstreams as they want to see things or how they wish to control traffic to and from their upstreams into their networks.

yeh, but compare the telstraclear and odyssey bgp paths.

odyssey: 4610 4739 20473

telstraclear: 9901 4637 1221 7474 4739 4739 20473

it's not like telstraclear are actually advertising a short path.

if reach or telstra peered at equnix sydney it'd be fine. or if snap had a way to peer at equinix sydney.

it's like telstraclear sending data from wellington to vodafone wellington through auckland, because telstraclear won't peer at wix in wellington. when most large isp's in this country are peering at wix. (i dunno if htat actually happens or not)

Ragnor

8221 posts

Uber Geek

Trusted

#609470 15-Apr-2012 03:44

Undoubtedly Telstraclear/Telstra/Reach run the numbers periodically and find that benefit from openly peering at Equnix doesn't outweigh the revenue they would lose from transit customers who wouldn't need transit from them any more.

Boo hoo!

If you care about this issue vote with your wallet and change ISP.