Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Cisco: NSP

GRE tunnel bandwidth

 

 

Cisco nsp RSS feed   Index | Next | Previous | View Threaded


jneiberger at gmail

Aug 4, 2012, 8:57 AM

Post #1 of 5 (2798 views)
Permalink
GRE tunnel bandwidth

I have some users experiencing slow file transfers over a GRE tunnel. The
tunnel is riding over 10-gig links. I see that the default tunnel bandwidth
is 8 Mbps. Does that mean that the tunnel is rate limited to that value? If
so, is the simple solution raising the bandwidth with the "tunnel bandwidth
transmit" command?

Thanks,
John
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


chuckchurch at gmail

Aug 4, 2012, 9:20 AM

Post #2 of 5 (2708 views)
Permalink
Re: GRE tunnel bandwidth [In reply to]

Tunnel bandwidth command (or any interface bandwidth) is used for
statistics-computation only. It does factor into QOS too if you use
percentage type commands. I'm guessing there are two possible things to
look at. The CPU of the devices doing the tunnel endpoints is high because
of the encapsulation, or else the tunnel MTU is affecting the clients (if
TCP).

Chuck

-----Original Message-----
From: cisco-nsp-bounces [at] puck
[mailto:cisco-nsp-bounces [at] puck] On Behalf Of John Neiberger
Sent: Saturday, August 04, 2012 11:57 AM
To: cisco-nsp [at] puck
Subject: [c-nsp] GRE tunnel bandwidth

I have some users experiencing slow file transfers over a GRE tunnel. The
tunnel is riding over 10-gig links. I see that the default tunnel bandwidth
is 8 Mbps. Does that mean that the tunnel is rate limited to that value? If
so, is the simple solution raising the bandwidth with the "tunnel bandwidth
transmit" command?

Thanks,
John
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jneiberger at gmail

Aug 4, 2012, 9:26 AM

Post #3 of 5 (2716 views)
Permalink
Re: GRE tunnel bandwidth [In reply to]

Thanks, man. It was difficult to tell what that command did from the
command reference. It simply says that it sets the bandwidth used to
transmit packets. That sounds like more than just a statistical
command, but I think you're correct.

The endpoints are 7600s with Sup 720. The MTU in the path is over 9000
and the tunnel MTU is lower than that, but I think we should try
lowering the tunnel MTU a bit and possibly use tcp mss adjust. I don't
know that it's necessary, though. I believe the end devices themselves
have an MTU of 1500, so we should have plenty of room in the tunnel
for those packets. I'll dig a little deeper into it.

Thanks!
John


On Sat, Aug 4, 2012 at 10:20 AM, Chuck Church <chuckchurch [at] gmail> wrote:
> Tunnel bandwidth command (or any interface bandwidth) is used for
> statistics-computation only. It does factor into QOS too if you use
> percentage type commands. I'm guessing there are two possible things to
> look at. The CPU of the devices doing the tunnel endpoints is high because
> of the encapsulation, or else the tunnel MTU is affecting the clients (if
> TCP).
>
> Chuck
>
> -----Original Message-----
> From: cisco-nsp-bounces [at] puck
> [mailto:cisco-nsp-bounces [at] puck] On Behalf Of John Neiberger
> Sent: Saturday, August 04, 2012 11:57 AM
> To: cisco-nsp [at] puck
> Subject: [c-nsp] GRE tunnel bandwidth
>
> I have some users experiencing slow file transfers over a GRE tunnel. The
> tunnel is riding over 10-gig links. I see that the default tunnel bandwidth
> is 8 Mbps. Does that mean that the tunnel is rate limited to that value? If
> so, is the simple solution raising the bandwidth with the "tunnel bandwidth
> transmit" command?
>
> Thanks,
> John
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


ler762 at gmail

Aug 4, 2012, 9:48 AM

Post #4 of 5 (2722 views)
Permalink
Re: GRE tunnel bandwidth [In reply to]

On 8/4/12, Chuck Church <chuckchurch [at] gmail> wrote:
> Tunnel bandwidth command (or any interface bandwidth) is used for
> statistics-computation only. It does factor into QOS too if you use
> percentage type commands.

It's also used for eigrp route metrics.

I suspect routing isn't the issue here tho
> > I have some users experiencing slow file transfers over a GRE tunnel. The
> > tunnel is riding over 10-gig links.

To be able to use all the bandwidth, the tcp window size needs to be
at least <bandwidth in bytes/sec> * <round trip time in seconds>
So, for example, on a 10Gb link with a 2ms round trip time the tcp
window needs to be 2.5MB. I remember WinXP defaulting to 16KB...

Regards,
Lee


> I'm guessing there are two possible things to
> look at. The CPU of the devices doing the tunnel endpoints is high because
> of the encapsulation, or else the tunnel MTU is affecting the clients (if
> TCP).
>
> Chuck
>
> -----Original Message-----
> From: cisco-nsp-bounces [at] puck
> [mailto:cisco-nsp-bounces [at] puck] On Behalf Of John Neiberger
> Sent: Saturday, August 04, 2012 11:57 AM
> To: cisco-nsp [at] puck
> Subject: [c-nsp] GRE tunnel bandwidth
>
> I have some users experiencing slow file transfers over a GRE tunnel. The
> tunnel is riding over 10-gig links. I see that the default tunnel bandwidth
> is 8 Mbps. Does that mean that the tunnel is rate limited to that value? If
> so, is the simple solution raising the bandwidth with the "tunnel bandwidth
> transmit" command?
>
> Thanks,
> John
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jneiberger at gmail

Aug 4, 2012, 11:02 AM

Post #5 of 5 (2727 views)
Permalink
Re: GRE tunnel bandwidth [In reply to]

I am having them check their window size settings, but I just found an
MTU problem in the path. They're using jumbo frames, but the next hop
in the path after the tunnel source had an MTU of 1500. I've corrected
that and I bet they'll see a lot better performance.

Thanks!
John

On Sat, Aug 4, 2012 at 10:48 AM, Lee <ler762 [at] gmail> wrote:
> On 8/4/12, Chuck Church <chuckchurch [at] gmail> wrote:
>> Tunnel bandwidth command (or any interface bandwidth) is used for
>> statistics-computation only. It does factor into QOS too if you use
>> percentage type commands.
>
> It's also used for eigrp route metrics.
>
> I suspect routing isn't the issue here tho
>> > I have some users experiencing slow file transfers over a GRE tunnel. The
>> > tunnel is riding over 10-gig links.
>
> To be able to use all the bandwidth, the tcp window size needs to be
> at least <bandwidth in bytes/sec> * <round trip time in seconds>
> So, for example, on a 10Gb link with a 2ms round trip time the tcp
> window needs to be 2.5MB. I remember WinXP defaulting to 16KB...
>
> Regards,
> Lee
>
>
>> I'm guessing there are two possible things to
>> look at. The CPU of the devices doing the tunnel endpoints is high because
>> of the encapsulation, or else the tunnel MTU is affecting the clients (if
>> TCP).
>>
>> Chuck
>>
>> -----Original Message-----
>> From: cisco-nsp-bounces [at] puck
>> [mailto:cisco-nsp-bounces [at] puck] On Behalf Of John Neiberger
>> Sent: Saturday, August 04, 2012 11:57 AM
>> To: cisco-nsp [at] puck
>> Subject: [c-nsp] GRE tunnel bandwidth
>>
>> I have some users experiencing slow file transfers over a GRE tunnel. The
>> tunnel is riding over 10-gig links. I see that the default tunnel bandwidth
>> is 8 Mbps. Does that mean that the tunnel is rate limited to that value? If
>> so, is the simple solution raising the bandwidth with the "tunnel bandwidth
>> transmit" command?
>>
>> Thanks,
>> John
>> _______________________________________________
>> cisco-nsp mailing list cisco-nsp [at] puck
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
>> _______________________________________________
>> cisco-nsp mailing list cisco-nsp [at] puck
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Cisco nsp RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.