Print

[Ninho]

How is the recommended PoP chosen on the subscription page ? Using a table for IP geolocation, or is it somehow testing for real latency ?

I'm asking because the page offered me a PoP situated in another country, which I declined in favour of a closer location, geographically. Maybe I shouldn't have overridden the automatic choice ...


In the remotely possible event I switched servers, would I keep the same IP6 numbers, or are they pooled per HE site ?

[kriteknetworks]

If you change tunnel servers, you lose the old IPs, and get new ones from the new tserv.

[Ninho]

Thanks Kriteknetworks ! I'd need a very good reason to change then.
Someone can tell how the system works out the determination of a default PoP ?

And/or tell me what the IP(v4) of the Frankfürt server is, that I can make ping/traceoute/timing tests from my location.

[piojan]

That geolocation work on the bases of anycasting.
It's the closes tserv to the place you ips dropes traffic to HE.

There might be some issues (like you isp not peering with he at certain places).
So network wise this is the closes server.

As for the ip:
tserv6.fra1.ipv6.he.net
(that host name has both A and AAAA record - ignore the last one)

[broquea]

Did you know that on the tunnel creation page, all the IPs are listed right next to the tunnel-server locations? It's true!

So the auto-recommendation basically is determined by your closest path to a specific IPv4 address.

[Ninho]

Did you know that on the tunnel creation page, all the IPs are listed right next to the tunnel-server locations? It's true!

Oh they're hidden in full view ! Strange case of blindness 

So the auto-recommendation basically is determined by your closest path to a specific IPv4 address.

I shall trace the routes to determine if the difference in worth changing servers.
Shouldn't you consider allocating from a unique pool and having a global database shared between the servers sometime in the future ? It will allow easy fall over in case of one PoP being momentarily out of service.

[kcochran]

Shouldn't you consider allocating from a unique pool and having a global database shared between the servers sometime in the future ? It will allow easy fall over in case of one PoP being momentarily out of service.

The sort of thing that would do to the routing table is incredibly ugly.  There are a number of IPv4 networks which did that in their youth, and they're now paying for it with significantly increased routing complexity.

[piojan]

I shall trace the routes to determine if the difference in worth changing servers.

The ip that you have exidently provided belonged to opentransit/francetelecom.

Looks like the problem is that Telia dosn't use (or it dosn't exist) the peering with HE at paris.

Code: [Select]

Tracing the route to 74.82.42.43
  1 tengige0-15-0-1.auvtr1.Aubervilliers.opentransit.net (193.251.241.181) 4 msec 0 msec 0 msec
  2 pos0-6-1-0.pastr1.Paris.opentransit.net (193.251.243.30) 0 msec 0 msec 0 msec
  3 telia-4.GW.opentransit.net (193.251.254.154) 4 msec 4 msec 0 msec
  4 prs-bb1-link.telia.net (80.91.253.109) [AS 1299] 0 msec
    prs-bb2-link.telia.net (80.91.248.10) [AS 1299] 0 msec
    prs-bb1-link.telia.net (80.91.250.53) [AS 1299] 0 msec
  5 ffm-bb2-link.telia.net (80.91.252.234) [AS 1299] 8 msec
    ffm-bb1-link.telia.net (80.91.251.211) [AS 1299] 8 msec
    ffm-bb2-link.telia.net (80.91.248.66) [AS 1299] 8 msec
  6 ffm-b2-link.telia.net (80.91.249.103) [AS 1299] 12 msec
    ffm-b2-link.telia.net (80.91.252.174) [AS 1299] 12 msec
    ffm-b2-link.telia.net (80.91.248.98) [AS 1299] 12 msec
  7 hurricane-ic-129711-ffm-b2.c.telia.net (213.248.92.34) [AS 1299] 12 msec 12 msec 12 msec
  8 74.82.42.43 [AS 6939] 8 msec 12 msec 12 msec


[Ninho]

The ip that you have provided belonged to opentransit/francetelecom.
Looks like the problem is that Telia dosn't use (or it dosn't exist) the peering with HE at paris.

Right, I'm a France-Telecom/Orange residential customer.
I don't know zilch about peering, but you must be right again : my traceroutes to HE servers in Frankfurt, Amsterdam and London are approx. 75, 76 and 77 ms, versus Paris : 80 ms.

My bad for trying to outsmart the system. Never mind, I'll be wiser when it's time to request a new tunnel. 

Code: [Select]

1 tengige0-15-0-1.auvtr1.Aubervilliers.opentransit.net (193.251.241.181) 4 msec 0 msec 0 msec

You must be living in the transit centre, Piojan 

[jimb]

It's also nice living about 10 miles from one of the Fremont tunnel servers.  I get about 8.7MS avg.   

[piojan]

Code: [Select]

1 tengige0-15-0-1.auvtr1.Aubervilliers.opentransit.net (193.251.241.181) 4 msec 0 msec 0 msec

You must be living in the transit centre, Piojan 

I wish

That traceroue was taken form the OT looking glass.

I personaly have 36 ms to the nearest tserv. (Europien side of the globe).