Concerning MPLS paths

Marshall Eubanks tme at americafree.tv
Tue Apr 28 04:13:05 CDT 2009


On Apr 28, 2009, at 4:51 AM, Saqib Ilyas wrote:

> Anyone?
>
>
> On Mon, Apr 27, 2009 at 6:15 PM, Saqib Ilyas <msaqib at gmail.com> wrote:
>
>> Furthermore, I was also wondering, if the bandwidth constraints are  
>> upper
>> bounds, what does the traffic distribution typically look like at  
>> an LSR?

It clearly depends on what that traffic is.

The MPLS services I am most familiar with carry video traffic, with  
traffic patterns that look very different from the typical web site  
(generally the traffic is either on or off, there is very little  
"burstiness," there can be long periods of basically full usage of the  
available bandwidth and, if you commit to X Mbps, you had better  
actually have it, not X - epsilon). I would guess that this is one end  
of the spectrum, that bursty web traffic is the other, and that most  
other uses (such as VOIP) fall somewhere in between.

Regards
Marshall


>>
>> We're interested in traffic within a single service provider, non- 
>> Internet
>> traffic. Perhaps most service providers set aside some (dynamic?)  
>> pool for
>> Internet traffic, while making commitments to customer's inter-site  
>> traffic.
>> Thanks and best regards

>>
>>
>>
>> On Mon, Apr 27, 2009 at 5:50 PM, Saqib Ilyas <msaqib at gmail.com>  
>> wrote:
>>
>>> William
>>> Thanks for the reply. You say that LSPs are not static unless you  
>>> use TE
>>> tunnels. Are you referring to the staticness in terms of the path  
>>> or in the
>>> amount of bandwidth reserved on each link along a fixed path  
>>> determined at
>>> the time of signalling? Isn't a bandwidth constrained LSP always a  
>>> TE
>>> tunnel?
>>> Thanks and best regards
>>>
>>>
>>> On Mon, Apr 27, 2009 at 5:41 PM, William McCall <william.mccall at gmail.com
>>>> wrote:
>>>
>>>> Well, yes (if you don't count the additional traffic of
>>>> signalling/routing protocols, label imposition, etc) but consider  
>>>> the fact
>>>> that topologies change and routing will tend to change the total  
>>>> traffic
>>>> handled through a node. LSPs are not static unless you use TE  
>>>> tunnels.
>>>> Remember that labels are Forwarding Equivalency Classes and that  
>>>> translates
>>>> into subnets (whether they're subnets in a L3 vpn or part of the  
>>>> P network)
>>>> and the routing is still handled through an IGP or BGP.
>>>>
>>>> HTH
>>>>
>>>> --WJM IV
>>>>
>>>>
>>>> On Mon, Apr 27, 2009 at 7:10 AM, Saqib Ilyas <msaqib at gmail.com>  
>>>> wrote:
>>>>
>>>>> Hello everyone
>>>>> In the context of a single service provider network running  
>>>>> MPLS, if a
>>>>> number of bandwidth constrained LSPs are passing through a  
>>>>> particular
>>>>> node
>>>>> and the sum of the bandwidth constraints for the LSPs is X Mb/s,  
>>>>> then is
>>>>> X
>>>>> the upper bound on the traffic through that node, or is it  
>>>>> sometimes
>>>>> exceeded as well?
>>>>> Thanks and best regards
>>>>>
>>>>
>>>>
>>>
>>>
>>> --
>>> Muhammad Saqib Ilyas
>>> PhD Student, Computer Science and Engineering
>>> Lahore University of Management Sciences
>>>
>>
>>
>>
>> --
>> Muhammad Saqib Ilyas
>> PhD Student, Computer Science and Engineering
>> Lahore University of Management Sciences
>>
>
>
>
> -- 
> Muhammad Saqib Ilyas
> PhD Student, Computer Science and Engineering
> Lahore University of Management Sciences
>

Regards
Marshall Eubanks
CEO / AmericaFree.TV





More information about the NANOG mailing list