Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.networking > #4551 > unrolled thread

routing

Back to article view | Back to comp.os.linux.networking

Contents

  routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-01 15:53 +0000
    Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-01 18:31 +0200
      Re: routing William Unruh <unruh@invalid.ca> - 2020-06-01 22:30 +0000
        Re: routing "Carlos E. R." <robin_listas@es.invalid> - 2020-06-02 14:50 +0200
          Re: routing William Unruh <unruh@invalid.ca> - 2020-06-02 19:54 +0000
      Re: routing Ken Hart <kwhart1@frontier.com> - 2020-06-01 18:35 -0400
        Re: routing "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2020-06-01 19:00 -0400
          Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-02 02:18 +0200
            Re: routing "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2020-06-01 20:32 -0400
              Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-02 08:12 +0200
                Re: routing "Carlos E. R." <robin_listas@es.invalid> - 2020-06-02 14:54 +0200
                  Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-02 15:53 +0200
                    Re: routing "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2020-06-02 11:19 -0400
                    Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-02 21:25 +0200
                    Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:44 -0700
                      Re: routing Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2020-06-05 17:31 +0000
                        Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 09:20 -0700
                    Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:46 -0700
                Re: routing "David W. Hodgins" <dwhodgins@nomail.afraid.org> - 2020-06-02 09:38 -0400
                  Re: routing Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2020-06-02 16:21 +0000
                    Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-02 21:15 +0200
                    Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-02 21:36 +0200
                  Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:48 -0700
                    Re: routing William Unruh <unruh@invalid.ca> - 2020-06-05 04:54 +0000
            Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-02 21:30 +0200
              Re: routing Tauno Voipio <tauno.voipio@notused.fi.invalid> - 2020-06-03 13:51 +0300
                Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-03 13:30 +0200
              Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:51 -0700
                Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-05 15:23 +0200
                  Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:19 -0700
                    Re: routing William Unruh <unruh@invalid.ca> - 2020-06-10 17:44 +0000
            Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:41 -0700
              Re: routing William Unruh <unruh@invalid.ca> - 2020-06-05 04:50 +0000
              Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-05 09:44 +0200
                Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:36 -0700
          Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:40 -0700
            Re: routing William Unruh <unruh@invalid.ca> - 2020-06-05 04:47 +0000
              Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-05 09:53 +0200
              Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:41 -0700
                Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-10 12:40 +0200
                  Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 09:25 -0700
                    Re: routing Dan Purgert <dan@djph.net> - 2020-06-12 09:45 +0000
                    Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-12 13:55 +0200
                      Re: routing David Brown <david.brown@hesbynett.no> - 2020-06-12 15:45 +0200
                        Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-12 23:21 +0200
                      Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-12 22:55 -0700
            Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-05 15:34 +0200
              Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:45 -0700
                Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-10 12:52 +0200
      Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-04 20:35 -0700
        Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-05 15:38 +0200
          Re: routing Aragorn <thorongil@telenet.be> - 2020-06-05 15:46 +0200
            Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-06 00:00 +0200
          Re: routing Dan Purgert <dan@djph.net> - 2020-06-09 18:57 +0000
            Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-09 21:08 +0200
              Re: routing Dan Purgert <dan@djph.net> - 2020-06-09 22:07 +0000
                Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-10 00:36 +0200
          Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:48 -0700
        Re: routing Aragorn <thorongil@telenet.be> - 2020-06-05 15:45 +0200
    Re: routing William Unruh <unruh@invalid.ca> - 2020-06-01 22:21 +0000
      Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-02 02:39 +0000
        Re: routing William Unruh <unruh@invalid.ca> - 2020-06-02 04:30 +0000
          Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-02 12:03 +0000
    Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-02 20:34 +0000
      Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-02 23:08 +0200
      Re: routing William Unruh <unruh@invalid.ca> - 2020-06-02 21:20 +0000
    Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-05 15:25 +0000
      Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-06 00:02 +0200
        Re: routing Jorgen Grahn <grahn+nntp@snipabacken.se> - 2020-06-06 08:04 +0000
          Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-06 12:00 +0200
            Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-06 10:43 +0000
              Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-06 13:35 +0200
                Re: routing Bob Tennent <rdtennent@tennent.ca> - 2020-06-06 12:04 +0000
                  Re: routing "Carlos E.R." <robin_listas@es.invalid> - 2020-06-06 20:35 +0200
            Re: routing Johann Beretta <beretta@nun-ya-bizness.com> - 2020-06-10 00:55 -0700
          Re: routing William Unruh <unruh@invalid.ca> - 2020-06-06 16:19 +0000

Page 4 of 4 — ← Prev page 1 2 3 [4]

#4559

On Mon, 1 Jun 2020 22:21:03 -0000 (UTC), William Unruh wrote:

 > On 2020-06-01, Bob Tennent <rdtennent@tennent.ca> wrote:
 >> I need a low-latency connection to a colleague in the
 >> same city (Kingston Ont.) but who has a different ISP.
 >> I've traced the route of packets and they go via Toronto,
 >> Chicago, Buffalo, and finally back to Kingston (!). I've
 >> asked tech support for my ISP and they say they can't do
 >> anything about this. Is there anything I can do on my
 >> system, such as routing packets directly to the final
 >> destination or to the Buffalo site?

 > Why do you care? What is the latency you are getting? And
 > you could always both get the same ISP, who might direct
 > the packets more directly.

A group of us are trying to use Jamulus:

http://llcon.sourceforge.net/

The server is on Bell. Those on Bell are getting < 40 ms
latency; but those who are on Cogeco are getting > 40 ms.

[toc] | [prev] | [next] | [standalone]

#4560

On 2020-06-02, Bob Tennent <rdtennent@tennent.ca> wrote:
> On Mon, 1 Jun 2020 22:21:03 -0000 (UTC), William Unruh wrote:
>
> > On 2020-06-01, Bob Tennent <rdtennent@tennent.ca> wrote:
> >> I need a low-latency connection to a colleague in the
> >> same city (Kingston Ont.) but who has a different ISP.
> >> I've traced the route of packets and they go via Toronto,
> >> Chicago, Buffalo, and finally back to Kingston (!). I've
> >> asked tech support for my ISP and they say they can't do
> >> anything about this. Is there anything I can do on my
> >> system, such as routing packets directly to the final
> >> destination or to the Buffalo site?
>
> > Why do you care? What is the latency you are getting? And
> > you could always both get the same ISP, who might direct
> > the packets more directly.
>
> A group of us are trying to use Jamulus:
>
> http://llcon.sourceforge.net/
>
> The server is on Bell. Those on Bell are getting < 40 ms
> latency; but those who are on Cogeco are getting > 40 ms.

Light goes at 300000 km/sec. So from one place in Kingston to another
should take onely a few microseconds. Thus both the Bell and Cogeco are
taking roundabout routes to get there. I am sort of surprized that the
inter-carrier route is taking such a roundabout route.

The internet is not designed for low latency. I can see where prople
trying to play music together would find high latency problematic. You
just have to all become organists which instruments has latency in the
bass of almost a second. Or get yourselves all on the same network.
Also sometimes systems will do packet coalescence,-- ie wait until you
have lots of packets and sent them all at once. Thus while the network
latency is small, waiting for all the packets to be collected could add
lots of additional latency (this is a problem for example for ntp time
synchronisation).

[toc] | [prev] | [next] | [standalone]

#4562

On Tue, 2 Jun 2020 04:30:43 -0000 (UTC), William Unruh wrote:
 > On 2020-06-02, Bob Tennent <rdtennent@tennent.ca> wrote:
 >> On Mon, 1 Jun 2020 22:21:03 -0000 (UTC), William Unruh wrote:
 >>
 >> > On 2020-06-01, Bob Tennent <rdtennent@tennent.ca> wrote:
 >> >> I need a low-latency connection to a colleague in the
 >> >> same city (Kingston Ont.) but who has a different ISP.
 >> >> I've traced the route of packets and they go via Toronto,
 >> >> Chicago, Buffalo, and finally back to Kingston (!). I've
 >> >> asked tech support for my ISP and they say they can't do
 >> >> anything about this. Is there anything I can do on my
 >> >> system, such as routing packets directly to the final
 >> >> destination or to the Buffalo site?
 >>
 >> > Why do you care? What is the latency you are getting? And
 >> > you could always both get the same ISP, who might direct
 >> > the packets more directly.
 >>
 >> A group of us are trying to use Jamulus:
 >>
 >> http://llcon.sourceforge.net/
 >>
 >> The server is on Bell. Those on Bell are getting < 40 ms
 >> latency; but those who are on Cogeco are getting > 40 ms.
 >
 > Light goes at 300000 km/sec. So from one place in
 > Kingston to another should take onely a few microseconds.
 > Thus both the Bell and Cogeco are taking roundabout
 > routes to get there. I am sort of surprized that the
 > inter-carrier route is taking such a roundabout route.

Presumably intended to optimize streaming access to Netflix
etc.

 > The internet is not designed for low latency. I can see
 > where prople trying to play music together would find
 > high latency problematic.

The Jamulus software can cope with < 40 ms latency. It
merges inputs from several "clients", synchronizes them and
returns a group "performance" that lags slightly behind
everyone's playing, but this is tolerable if you use
headphones to listen to the merged output and not to your
own playing.

It may be that the > 40 ms latency in the Bell-to-Cogeco
route will work; we're still trying to cope with various
technical issues (good microphone, good analogue-to-digital
conversion, ethernet connection to a Mac box that doesn't
have an ethernet port, etc.)

[toc] | [prev] | [next] | [standalone]

#4574

On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
 > I need a low-latency connection to a colleague in the
 > same city (Kingston Ont.) but who has a different ISP.
 > I've traced the route of packets and they go via Toronto,
 > Chicago, Buffalo, and finally back to Kingston (!). I've
 > asked tech support for my ISP and they say they can't do
 > anything about this. Is there anything I can do on my
 > system, such as routing packets directly to the final
 > destination or to the Buffalo site?

So it seems changing the routing between the ISPs is not feasible.

How about this: suppose I get internet from *both* ISPs.
How would I route incoming packets from a computer on one
network to a computer on the LAN of the other network? In
other words, re-route *internally*. To reduce latency, this
should all be done with wired (ethernet) connections. Can
one even buy a non-wifi router these days?

[toc] | [prev] | [next] | [standalone]

#4575

On 02/06/2020 22.34, Bob Tennent wrote:
> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>   > I need a low-latency connection to a colleague in the
>   > same city (Kingston Ont.) but who has a different ISP.
>   > I've traced the route of packets and they go via Toronto,
>   > Chicago, Buffalo, and finally back to Kingston (!). I've
>   > asked tech support for my ISP and they say they can't do
>   > anything about this. Is there anything I can do on my
>   > system, such as routing packets directly to the final
>   > destination or to the Buffalo site?
> 
> So it seems changing the routing between the ISPs is not feasible.
> 
> How about this: suppose I get internet from *both* ISPs.
> How would I route incoming packets from a computer on one
> network to a computer on the LAN of the other network? In
> other words, re-route *internally*. 

Basically, you change the routing table on your computer. Command "route".


> To reduce latency, this
> should all be done with wired (ethernet) connections. Can
> one even buy a non-wifi router these days?

Certainly.


-- 
Cheers, Carlos.

[toc] | [prev] | [next] | [standalone]

#4576

On 2020-06-02, Bob Tennent <rdtennent@tennent.ca> wrote:
> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
> > I need a low-latency connection to a colleague in the
> > same city (Kingston Ont.) but who has a different ISP.
> > I've traced the route of packets and they go via Toronto,
> > Chicago, Buffalo, and finally back to Kingston (!). I've
> > asked tech support for my ISP and they say they can't do
> > anything about this. Is there anything I can do on my
> > system, such as routing packets directly to the final
> > destination or to the Buffalo site?
>
> So it seems changing the routing between the ISPs is not feasible.
>
> How about this: suppose I get internet from *both* ISPs.
> How would I route incoming packets from a computer on one

Put 2 network cards into the computer and plug on router into one card
and the other into the other. They will have different IP addresses for
the interfaces. Now I do not know what you want to route from one
interface to the other.

> network to a computer on the LAN of the other network? In
> other words, re-route *internally*. To reduce latency, this
> should all be done with wired (ethernet) connections. Can
> one even buy a non-wifi router these days?

Sure, and besides the wifi routers usually come with a bunch of ethernet
ports which you can use.

[toc] | [prev] | [next] | [standalone]

#4596

On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
 > I need a low-latency connection to a colleague in the
 > same city (Kingston Ont.) but who has a different ISP.
 > I've traced the route of packets and they go via Toronto,
 > Chicago, Buffalo, and finally back to Kingston (!). I've
 > asked tech support for my ISP and they say they can't do
 > anything about this. Is there anything I can do on my
 > system, such as routing packets directly to the final
 > destination or to the Buffalo site?

Thanks to everyone for their comments. I'm sorry I didn't
specify that Kingston Ontario is in Canada, but it's surely
the number of hops rather than the distance that is crucial
for latency.

I think I have a solution to my problem, albeit expensive:
if I subscribe to *both* ISPs using two ethernet interfaces
to the server, I believe clients could reach the server
software directly without the "detours". I'm less sure that
the server (Jamulus) is clever enough to route outgoing
packets to the right interface but perhaps I could configure
routes for specific IP addresses.

Is this going to work?

[toc] | [prev] | [next] | [standalone]

#4599

On 05/06/2020 17.25, Bob Tennent wrote:
> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>   > I need a low-latency connection to a colleague in the
>   > same city (Kingston Ont.) but who has a different ISP.
>   > I've traced the route of packets and they go via Toronto,
>   > Chicago, Buffalo, and finally back to Kingston (!). I've
>   > asked tech support for my ISP and they say they can't do
>   > anything about this. Is there anything I can do on my
>   > system, such as routing packets directly to the final
>   > destination or to the Buffalo site?
> 
> Thanks to everyone for their comments. I'm sorry I didn't
> specify that Kingston Ontario is in Canada, but it's surely
> the number of hops rather than the distance that is crucial
> for latency.
> 
> I think I have a solution to my problem, albeit expensive:
> if I subscribe to *both* ISPs using two ethernet interfaces
> to the server, I believe clients could reach the server
> software directly without the "detours". I'm less sure that
> the server (Jamulus) is clever enough to route outgoing
> packets to the right interface but perhaps I could configure
> routes for specific IP addresses.
> 
> Is this going to work?
> 

A router should automatically find the "less expensive" route.

But yes, each ISP has a range of IPs assigned. Finding them all would 
not be trivial, though.

-- 
Cheers, Carlos.

[toc] | [prev] | [next] | [standalone]

#4600

On Fri, 2020-06-05, Carlos E.R. wrote:
> On 05/06/2020 17.25, Bob Tennent wrote:
>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>>   > I need a low-latency connection to a colleague in the
>>   > same city (Kingston Ont.) but who has a different ISP.
>>   > I've traced the route of packets and they go via Toronto,
>>   > Chicago, Buffalo, and finally back to Kingston (!). I've
>>   > asked tech support for my ISP and they say they can't do
>>   > anything about this. Is there anything I can do on my
>>   > system, such as routing packets directly to the final
>>   > destination or to the Buffalo site?
>> 
>> Thanks to everyone for their comments. I'm sorry I didn't
>> specify that Kingston Ontario is in Canada, but it's surely
>> the number of hops rather than the distance that is crucial
>> for latency.
>> 
>> I think I have a solution to my problem, albeit expensive:
>> if I subscribe to *both* ISPs using two ethernet interfaces
>> to the server, I believe clients could reach the server
>> software directly without the "detours". I'm less sure that
>> the server (Jamulus) is clever enough to route outgoing
>> packets to the right interface but perhaps I could configure
>> routes for specific IP addresses.
>> 
>> Is this going to work?
>> 
>
> A router should automatically find the "less expensive" route.

A real router which routes IP and understands dynamic routing
protocols, yes.  But one of those home "routers"?  That is less
clear.

/Jorgen

-- 
  // Jorgen Grahn <grahn@  Oo  o.   .     .
\X/     snipabacken.se>   O  o   .

[toc] | [prev] | [next] | [standalone]

#4601

On 06/06/2020 10.04, Jorgen Grahn wrote:
> On Fri, 2020-06-05, Carlos E.R. wrote:
>> On 05/06/2020 17.25, Bob Tennent wrote:
>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>>>    > I need a low-latency connection to a colleague in the
>>>    > same city (Kingston Ont.) but who has a different ISP.
>>>    > I've traced the route of packets and they go via Toronto,
>>>    > Chicago, Buffalo, and finally back to Kingston (!). I've
>>>    > asked tech support for my ISP and they say they can't do
>>>    > anything about this. Is there anything I can do on my
>>>    > system, such as routing packets directly to the final
>>>    > destination or to the Buffalo site?
>>>
>>> Thanks to everyone for their comments. I'm sorry I didn't
>>> specify that Kingston Ontario is in Canada, but it's surely
>>> the number of hops rather than the distance that is crucial
>>> for latency.
>>>
>>> I think I have a solution to my problem, albeit expensive:
>>> if I subscribe to *both* ISPs using two ethernet interfaces
>>> to the server, I believe clients could reach the server
>>> software directly without the "detours". I'm less sure that
>>> the server (Jamulus) is clever enough to route outgoing
>>> packets to the right interface but perhaps I could configure
>>> routes for specific IP addresses.
>>>
>>> Is this going to work?
>>>
>>
>> A router should automatically find the "less expensive" route.
> 
> A real router which routes IP and understands dynamic routing
> protocols, yes.  But one of those home "routers"?  That is less
> clear.

Yes, of course, "home routers" rarely have three routable ethernet ports 
- they have one "entry" port, and an exit row of switched (non "routed") 
ports.



-- 
Cheers, Carlos.

[toc] | [prev] | [next] | [standalone]

#4602

On Sat, 6 Jun 2020 12:00:52 +0200, Carlos E.R. wrote:
 > On 06/06/2020 10.04, Jorgen Grahn wrote:
 >> On Fri, 2020-06-05, Carlos E.R. wrote:
 >>> On 05/06/2020 17.25, Bob Tennent wrote:
 >>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
 >>>>    > I need a low-latency connection to a colleague in the
 >>>>    > same city (Kingston Ont.) but who has a different ISP.
 >>>>    > I've traced the route of packets and they go via Toronto,
 >>>>    > Chicago, Buffalo, and finally back to Kingston (!). I've
 >>>>    > asked tech support for my ISP and they say they can't do
 >>>>    > anything about this. Is there anything I can do on my
 >>>>    > system, such as routing packets directly to the final
 >>>>    > destination or to the Buffalo site?
 >>>>
 >>>> Thanks to everyone for their comments. I'm sorry I didn't
 >>>> specify that Kingston Ontario is in Canada, but it's surely
 >>>> the number of hops rather than the distance that is crucial
 >>>> for latency.
 >>>>
 >>>> I think I have a solution to my problem, albeit expensive:
 >>>> if I subscribe to *both* ISPs using two ethernet interfaces
 >>>> to the server, I believe clients could reach the server
 >>>> software directly without the "detours". I'm less sure that
 >>>> the server (Jamulus) is clever enough to route outgoing
 >>>> packets to the right interface but perhaps I could configure
 >>>> routes for specific IP addresses.
 >>>>
 >>>> Is this going to work?
 >>>>
 >>>
 >>> A router should automatically find the "less expensive" route.
 >> 
 >> A real router which routes IP and understands dynamic routing
 >> protocols, yes.  But one of those home "routers"?  That is less
 >> clear.
 >
 > Yes, of course, "home routers" rarely have three routable ethernet ports 
 > - they have one "entry" port, and an exit row of switched (non "routed") 
 > ports.

As I said, I'll be using two ethernet interfaces and so two routers.

 

[toc] | [prev] | [next] | [standalone]

#4603

On 06/06/2020 12.43, Bob Tennent wrote:
> On Sat, 6 Jun 2020 12:00:52 +0200, Carlos E.R. wrote:
>   > On 06/06/2020 10.04, Jorgen Grahn wrote:
>   >> On Fri, 2020-06-05, Carlos E.R. wrote:
>   >>> On 05/06/2020 17.25, Bob Tennent wrote:
>   >>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>   >>>>    > I need a low-latency connection to a colleague in the
>   >>>>    > same city (Kingston Ont.) but who has a different ISP.
>   >>>>    > I've traced the route of packets and they go via Toronto,
>   >>>>    > Chicago, Buffalo, and finally back to Kingston (!). I've
>   >>>>    > asked tech support for my ISP and they say they can't do
>   >>>>    > anything about this. Is there anything I can do on my
>   >>>>    > system, such as routing packets directly to the final
>   >>>>    > destination or to the Buffalo site?
>   >>>>
>   >>>> Thanks to everyone for their comments. I'm sorry I didn't
>   >>>> specify that Kingston Ontario is in Canada, but it's surely
>   >>>> the number of hops rather than the distance that is crucial
>   >>>> for latency.
>   >>>>
>   >>>> I think I have a solution to my problem, albeit expensive:
>   >>>> if I subscribe to *both* ISPs using two ethernet interfaces
>   >>>> to the server, I believe clients could reach the server
>   >>>> software directly without the "detours". I'm less sure that
>   >>>> the server (Jamulus) is clever enough to route outgoing
>   >>>> packets to the right interface but perhaps I could configure
>   >>>> routes for specific IP addresses.
>   >>>>
>   >>>> Is this going to work?
>   >>>>
>   >>>
>   >>> A router should automatically find the "less expensive" route.
>   >>
>   >> A real router which routes IP and understands dynamic routing
>   >> protocols, yes.  But one of those home "routers"?  That is less
>   >> clear.
>   >
>   > Yes, of course, "home routers" rarely have three routable ethernet ports
>   > - they have one "entry" port, and an exit row of switched (non "routed")
>   > ports.
> 
> As I said, I'll be using two ethernet interfaces and so two routers.

Not the same thing. Not what we are talking about.

You need to add support for routing protocols in your Linux machine for 
this to work.

Best option would be a single access router, with two "entry" ports, 
removing the ISP provided routers, or setting them as pass-through.

-- 
Cheers, Carlos.

[toc] | [prev] | [next] | [standalone]

#4604

On Sat, 6 Jun 2020 13:35:09 +0200, Carlos E.R. wrote:
 > On 06/06/2020 12.43, Bob Tennent wrote:
 >> On Sat, 6 Jun 2020 12:00:52 +0200, Carlos E.R. wrote:
 >>   > On 06/06/2020 10.04, Jorgen Grahn wrote:
 >>   >> On Fri, 2020-06-05, Carlos E.R. wrote:
 >>   >>> On 05/06/2020 17.25, Bob Tennent wrote:
 >>   >>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
 >>   >>>>    > I need a low-latency connection to a colleague in the
 >>   >>>>    > same city (Kingston Ont.) but who has a different ISP.
 >>   >>>>    > I've traced the route of packets and they go via Toronto,
 >>   >>>>    > Chicago, Buffalo, and finally back to Kingston (!). I've
 >>   >>>>    > asked tech support for my ISP and they say they can't do
 >>   >>>>    > anything about this. Is there anything I can do on my
 >>   >>>>    > system, such as routing packets directly to the final
 >>   >>>>    > destination or to the Buffalo site?
 >>   >>>>
 >>   >>>> Thanks to everyone for their comments. I'm sorry I didn't
 >>   >>>> specify that Kingston Ontario is in Canada, but it's surely
 >>   >>>> the number of hops rather than the distance that is crucial
 >>   >>>> for latency.
 >>   >>>>
 >>   >>>> I think I have a solution to my problem, albeit expensive:
 >>   >>>> if I subscribe to *both* ISPs using two ethernet interfaces
 >>   >>>> to the server, I believe clients could reach the server
 >>   >>>> software directly without the "detours". I'm less sure that
 >>   >>>> the server (Jamulus) is clever enough to route outgoing
 >>   >>>> packets to the right interface but perhaps I could configure
 >>   >>>> routes for specific IP addresses.
 >>   >>>>
 >>   >>>> Is this going to work?
 >>   >>>>
 >>   >>>
 >>   >>> A router should automatically find the "less expensive" route.
 >>   >>
 >>   >> A real router which routes IP and understands dynamic routing
 >>   >> protocols, yes.  But one of those home "routers"?  That is less
 >>   >> clear.
 >>   >
 >>   > Yes, of course, "home routers" rarely have three routable ethernet ports
 >>   > - they have one "entry" port, and an exit row of switched (non "routed")
 >>   > ports.
 >> 
 >> As I said, I'll be using two ethernet interfaces and so two routers.
 >
 > Not the same thing. Not what we are talking about.

As I'm the OP, you should be talking about what I'm talking about.
 
 > You need to add support for routing protocols in your Linux machine for 
 > this to work.
 >
 > Best option would be a single access router, with two "entry" ports, 
 > removing the ISP provided routers, or setting them as pass-through.

I already have two routers, neither provided by the ISPs.
Maybe not the "best" option but is it impossible?  

[toc] | [prev] | [next] | [standalone]

#4606

On 06/06/2020 14.04, Bob Tennent wrote:
> On Sat, 6 Jun 2020 13:35:09 +0200, Carlos E.R. wrote:
>   > On 06/06/2020 12.43, Bob Tennent wrote:
>   >> On Sat, 6 Jun 2020 12:00:52 +0200, Carlos E.R. wrote:
>   >>   > On 06/06/2020 10.04, Jorgen Grahn wrote:
>   >>   >> On Fri, 2020-06-05, Carlos E.R. wrote:
>   >>   >>> On 05/06/2020 17.25, Bob Tennent wrote:
>   >>   >>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>   >>   >>>>    > I need a low-latency connection to a colleague in the
>   >>   >>>>    > same city (Kingston Ont.) but who has a different ISP.
>   >>   >>>>    > I've traced the route of packets and they go via Toronto,
>   >>   >>>>    > Chicago, Buffalo, and finally back to Kingston (!). I've
>   >>   >>>>    > asked tech support for my ISP and they say they can't do
>   >>   >>>>    > anything about this. Is there anything I can do on my
>   >>   >>>>    > system, such as routing packets directly to the final
>   >>   >>>>    > destination or to the Buffalo site?
>   >>   >>>>
>   >>   >>>> Thanks to everyone for their comments. I'm sorry I didn't
>   >>   >>>> specify that Kingston Ontario is in Canada, but it's surely
>   >>   >>>> the number of hops rather than the distance that is crucial
>   >>   >>>> for latency.
>   >>   >>>>
>   >>   >>>> I think I have a solution to my problem, albeit expensive:
>   >>   >>>> if I subscribe to *both* ISPs using two ethernet interfaces
>   >>   >>>> to the server, I believe clients could reach the server
>   >>   >>>> software directly without the "detours". I'm less sure that
>   >>   >>>> the server (Jamulus) is clever enough to route outgoing
>   >>   >>>> packets to the right interface but perhaps I could configure
>   >>   >>>> routes for specific IP addresses.
>   >>   >>>>
>   >>   >>>> Is this going to work?
>   >>   >>>>
>   >>   >>>
>   >>   >>> A router should automatically find the "less expensive" route.
>   >>   >>
>   >>   >> A real router which routes IP and understands dynamic routing
>   >>   >> protocols, yes.  But one of those home "routers"?  That is less
>   >>   >> clear.
>   >>   >
>   >>   > Yes, of course, "home routers" rarely have three routable ethernet ports
>   >>   > - they have one "entry" port, and an exit row of switched (non "routed")
>   >>   > ports.
>   >>
>   >> As I said, I'll be using two ethernet interfaces and so two routers.
>   >
>   > Not the same thing. Not what we are talking about.
> 
> As I'm the OP, you should be talking about what I'm talking about.

No. We are proposing a procedure for you  and your problem, and you are 
not listening.

>   
>   > You need to add support for routing protocols in your Linux machine for
>   > this to work.
>   >
>   > Best option would be a single access router, with two "entry" ports,
>   > removing the ISP provided routers, or setting them as pass-through.
> 
> I already have two routers, neither provided by the ISPs.
> Maybe not the "best" option but is it impossible?

That way, how are going the two routers to negotiate the least expensive 
route? That's the job of a single router.

-- 
Cheers, Carlos.

[toc] | [prev] | [next] | [standalone]

#4617

[Multipart message — attachments visible in raw view] — view raw

On 6/6/20 3:00 AM, Carlos E.R. wrote:

> 
> Yes, of course, "home routers" rarely have three routable ethernet ports
> - they have one "entry" port, and an exit row of switched (non "routed")
> ports.
> 
> 
> 
True, but Mikrotik routers can have any number of ports configured as
WAN ports, so they may be able to accomplish that.. There's a bit more
information needed, that a home router is not going to have access to by
default, to start routing based on hops or distance in an natural
fashion. Especially when the gateway router for your ISP may be making
corrections to it's routing table with an extremely fine resolution
based on various factors like $$ cost, congestion, etc..

I'm fairly certain you could write a script to run a traceroute in a
kind of "dirty" way to periodically update your routing table to perhaps
gain a bit of performance..

[toc] | [prev] | [next] | [standalone]

#4605

On 2020-06-06, Jorgen Grahn <grahn+nntp@snipabacken.se> wrote:
> On Fri, 2020-06-05, Carlos E.R. wrote:
>> On 05/06/2020 17.25, Bob Tennent wrote:
>>> On Mon, 1 Jun 2020 15:53:00 -0000 (UTC), Bob Tennent wrote:
>>>   > I need a low-latency connection to a colleague in the
>>>   > same city (Kingston Ont.) but who has a different ISP.
>>>   > I've traced the route of packets and they go via Toronto,
>>>   > Chicago, Buffalo, and finally back to Kingston (!). I've
>>>   > asked tech support for my ISP and they say they can't do
>>>   > anything about this. Is there anything I can do on my
>>>   > system, such as routing packets directly to the final
>>>   > destination or to the Buffalo site?
>>> 
>>> Thanks to everyone for their comments. I'm sorry I didn't
>>> specify that Kingston Ontario is in Canada, but it's surely
>>> the number of hops rather than the distance that is crucial
>>> for latency.
>>> 
>>> I think I have a solution to my problem, albeit expensive:
>>> if I subscribe to *both* ISPs using two ethernet interfaces
>>> to the server, I believe clients could reach the server
>>> software directly without the "detours". I'm less sure that
>>> the server (Jamulus) is clever enough to route outgoing
>>> packets to the right interface but perhaps I could configure
>>> routes for specific IP addresses.
>>> 
>>> Is this going to work?
>>> 
>>
>> A router should automatically find the "less expensive" route.
>
> A real router which routes IP and understands dynamic routing
> protocols, yes.  But one of those home "routers"?  That is less
> clear.

AFAIK, a router's job is to send the packet on the next router who can
then decide where to send the packet, taking into account congestion,
time of travel, etc. But the ISP is going to direct the home packet to
their own concentrator site which may well be in the states, etc. Ie,
more considerations than just "fastest clear travel time" may well come
in to play.  And as I said, the next test 5 min later could well have
taken a very different route.
>

[toc] | [prev] | [standalone]

Page 4 of 4 — ← Prev page 1 2 3 [4]

Back to top | Article view | comp.os.linux.networking

csiph-web