Re: layer 2 fragmentation vs layer 3 fragmentation

Hi Shiran,

You're absolutely right about MSS, this is from RFC 879:

3. The TCP Maximum Segment Size Option

   TCP provides an option that may be used at the time a connection is
   established (only) to indicate the maximum size TCP segment that can
   be accepted on that connection. This Maximum Segment Size (MSS)
   announcement *(often mistakenly called a negotiation)* is sent from the
   data receiver to the data sender and says "I can accept TCP segments
   up to size X".

And regarding Window size, I didn't explain myself well probably. I didn't
want to say that the full Window size is sent without receiving an ACK but
if no ACK is received full window size will be sent anyway. And then once
RTO reached, it will start retransmission if there was no ACK.

Correct me if I'm wrong please.

Regards,

On Thu, Dec 23, 2010 at 8:12 AM, shiran guez <shiranp3_at_gmail.com> wrote:

> Although it is seem like TCP negotiate MSS and window size , it do not
> negotiate at all!
> each side send his own supported size for MSS and window telling the remote
> side what he is permitting him to send.
>
> And about window size and Ack relation that is also wrong, if an entire
> window would have been lost that is a lot of data, Ack is sent from
> the receiver side on every 1 MSS size of data, the sender retransmit in case
> a RTO retransmission time out expired and he did't got an Ack, the Window
> size is a buffer the max of data a unit can process before it get an Ack's
> on the full window (buffer) amount.
>
> On Wed, Dec 22, 2010 at 8:02 PM, George Goglidze <goglidze_at_gmail.com>wrote:
>
>> Hi,
>>
>> L2 Fragmentation:
>> Frame-Relay - FRF.12/FRF.11
>> MLPPP - Fragmentation and Interleaving
>>
>> And when TCP negotiates a Window size, that has nothing to do with
>> fragmentation.
>> That only means that one end is gonna send that window size before it'll
>> expect ACK from the other end.
>>
>> For the matter of MTU, TCP negotiates MSS size, which is by default 1460B.
>> That together with IP 20B and TCP 20B makes it 1500B.
>>
>> Regards,
>>
>> On Wed, Dec 22, 2010 at 6:37 PM, Sadiq Yakasai <sadiqtanko_at_gmail.com>
>> wrote:
>>
>> > I have not heard of L2 fragmentation. I know f L3 fragmentation. I also
>> > know
>> > about L4 segmentation though ;-)
>> >
>> > Let me give this a try, see if I remember my basics properly here:
>> >
>> > Your interfaces have configured MTU (by default or otherwise). Your
>> stack
>> > on
>> > the device learns this value and pushes it to the IP layer. When a user
>> > sends data (from an application, say a web-browser) that is less than
>> the
>> > MTU value, no fragmentation at L3 occurs. This data gets pushed down
>> from
>> > L7
>> > all the way down to the wire without any segmentation.
>> >
>> > Now, when the application data is above the configured MTU size, we
>> start
>> > at
>> > L4 (since its where the interest really is). L4 does its normal TCP
>> > windowing and negotiates a window size. Now, this is the L4
>> segmentation.
>> > The negotiated window size IIRC is a function of the buffer sizes of the
>> > application (or stacks) on the 2 ends of the connection (IP source and
>> > destination). Now the L4 packetizes the data and pushes down to the IP
>> > layer. The IP layer has already learnt about the MTU from the configured
>> > value on the interface. This causes fragmentation to occur (+/- the
>> header
>> > values, ofcourse). L3 hands the packet to L2 and the headers and field
>> are
>> > populated and the frame is put on the wire.
>> >
>> > Someone correct me if I have mixed any part of this please.
>> >
>> > Hope it helps abit.
>> >
>> > Sadiq
>> >
>> > On Wed, Dec 22, 2010 at 5:20 PM, HEMANTH RAJ <hemanthrj_at_gmail.com>
>> wrote:
>> >
>> > > Hey i have a little confusion
>> > > What is the difference between layer 2 fragmentation and Layer 3
>> > > fragmantation
>> > > And also tell me when L2 frag is used
>> > > and when L3 frag is used
>> > > can anyone share any gud documents regardin this
>> > >
>> > >
>> > > --
>> > > Problems arise because we talk,problems are not solved because we
>> don't
>> > > talk
>> > > So good or bad talk to your affectionate one's freely.
>> > >
>> > > Yours Friendly,
>> > > HP HEMANTH RAJ
>> > > Cisco Systems Inc.
>> > >
>> > >
>> > > Blogs and organic groups at http://www.ccie.net
>> > >
>> > >
>> _______________________________________________________________________
>> > > Subscription information may be found at:
>> > > http://www.groupstudy.com/list/CCIELab.html
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> > >
>> >
>> >
>> > --
>> > CCIEx2 (R&S|Sec) #19963
>> >
>> >
>> > Blogs and organic groups at http://www.ccie.net
>> >
>> > _______________________________________________________________________
>> > Subscription information may be found at:
>> > http://www.groupstudy.com/list/CCIELab.html
>>
>>
>> Blogs and organic groups at http://www.ccie.net
>>
>> _______________________________________________________________________
>> Subscription information may be found at:
>> http://www.groupstudy.com/list/CCIELab.html
>>
>>
>>
>>
>>
>>
>>
>>
>
>
> --
> Shiran Guez
> MCSE CCNP NCE1 JNCIA-ER CCIE #20572
> http://cciep3.blogspot.com
> http://www.linkedin.com/in/cciep3
> http://twitter.com/cciep3

Blogs and organic groups at http://www.ccie.net
Received on Thu Dec 23 2010 - 12:54:10 ART