[c-nsp] MTU Question on T1

[Joe Maimon]

Its already fragmented at 1500.....but still?

Richard J. Sears wrote:

>Hey Gert - 
>
>On the dual T1 issue - correct me if I am wrong please but with ppp
>multilink interleaving turned on, won't it take that single 40000 byte
>packet and shove it down one T1 because it came in as a single packet..?
>
>
>?
>
>
>Thanks
>
>
>On Mon, 24 Jan 2005 18:54:23 +0100
>Gert Doering <gert at greenie.muc.de> wrote:
>
>  
>
>>Hi,
>>
>>On Mon, Jan 24, 2005 at 09:21:56AM -0800, Richard J. Sears wrote:
>>    
>>
>>>Their entire claim is that because they cannot ping the other router
>>>with a 40000 byte packet with less than 50ms latency, that is the entire
>>>problem and until we fix that issue - don't call them back.
>>>
>>>So - my brain begins to think and I go to my 7500s, ping any number of
>>>customers with T1s with no traffic on them with Cisco's max packet size of
>>>18024 - I see around 225ms. 
>>>      
>>>
>>Pure physics will make this impossible.
>>
>>A T1 is 1,500,000 bits per second, which is about 190,000 bytes/second.
>>
>>If you send 40,000 bytes through that channel, just moving the bits one-way
>>will take 40/190 = 210 ms.  No fragmentation and reassembly involved,
>>just pure bits clocked at 1.5 Mbit/second.
>>
>>For the RTT, you have the data going back and forth, so the time will 
>>go up to 420 ms minimum.   Add actual latency to this (like "40ms" for
>>a longish T1).
>>
>>If your gear will distribute the packets 1:1 onto both T1s, you can 
>>speed up things a lot, because you can effectively pump with 3 Mbit/s.
>>- it won't be twice the rate due to more effort in reordering things,
>>but nearly so.
>>
>>In any case, there is no way to get a packet with 40 kbyte across a T1
>>in 50 ms.   So either the vendor in question is saying "our application
>>will not work over T1 links" (without admitting it), or they are just too
>>stupid to do the math.
>>
>>    
>>
>>>Each T1 is set with an mtu of 1500 bytes, which means that every 40000
>>>byte packet have to be broken up into almost 27 packets, get sent across
>>>the T1, be put back together and then reverse that process in order to
>>>get the RTT.
>>>
>>>I do not see any place to increase the MTU on a Cisco T1 WIC card (if
>>>even we would want to), nor do I see errors.
>>>      
>>>
>>Increasing the MTU won't help you there, as the packet is already
>>fragmented by the sending host.  Ethernet MTU is 1500...
>>
>>    
>>
>>>After looking at about 20 different T1s - I get the sense that I won't
>>>see anything less than 300 to 400ms doing a 40000 byte ping test across
>>>two Cisco 1720's.
>>>      
>>>
>>Correct conclusion.
>>
>>gert
>>-- 
>>USENET is *not* the non-clickable part of WWW!
>>                                                           //www.muc.de/~gert/
>>Gert Doering - Munich, Germany                             gert at greenie.muc.de
>>fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de
>>    
>>
>
>
>******************************************
>Richard J. Sears
>Vice President         
>American Internet Services                          
>----------------------------------------------------
>rsears at adnc.com
>http://www.adnc.com
>----------------------------------------------------
>858.576.4272 - Phone
>858.427.2401 - Fax
>INOC-DBA - 6130
>----------------------------------------------------
>
>I fly because it releases my mind 
>from the tyranny of petty things . . 
>
>
>"Work like you don't need the money, love like you've
>never been hurt and dance like you do when nobody's
>watching."
>
>_______________________________________________
>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>https://puck.nether.net/mailman/listinfo/cisco-nsp
>archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>
>  
>