The MPLS-OPS Archive

["Martin, Christian" <cmartin@gnilink.net>]

> Sorry if this is off topic but just so you guys know what it 
> was about 
> 
> The CE is an IBM 2216 acting as an APPN node with an FEP 
> which will only deliver and accept 4k packets :o(  (designed 
> for Token ring
> I think) 
> these packets are IP encapsulated SNA  so we can fragment 
> them in transit 
> as long as they get to the opposite CE as 4k packets once 
> again so we need
> to fragment and reintegrate 
> before passing to the CE  
> 
> Any good ideas ? 

Use DLSW+ between CEs.  This allows you to terminate the fragmented packets
with the Info frames intact.  Not sure about performance, though, or support
for that matter.  

regards,
chris



> 
> 
> Thanks again Eric and Gopal for your answers 
> 
> 
> 
> 
> -----Original Message-----
> From: Gopal@Home [mailto:netlanceconsulting@yahoo.com]
> Sent: 15 August 2001 13:11
> To: Ruyter Hill; raszuk@cisco.com; eosborne@cisco.com
> Cc: MPLS OPS (E-mail)
> Subject: Re: Packet fragmentation on MPLS VPN 
> 
> 
> Hi Hill, Till Robert or Eric answers.
> "will these be >re-integrated into a single packet at the 
> opposite (PE)
> before being passed >to the (CE) or will they be passed as 
> the fragments >to
> the CE"
> 
> ***packets will be fragmented at the the first device which has Mtu
> mismath(in this case PE) and un-fragmented at the 
> destination(the CE router
> or a PC in the C-network, whichever owns the destination ip addr)
> 
> ">>>not sure as I don't know when fragmentation occurs as the 
> initial label
> is >added at the VRF and the second at the outgoing interface 
> ( I think )
> >so in my understanding the packet would not be complete for 
> the VRF until
> it >was re-integrated at the opposite end >If this is the 
> case how will it
> effect PHP "
> 
> ***fragmentation wdn't effect the label operation. The first 
> device that
> fragments will use the same label for all the fragmented 
> pkts. The P LSR or
> the egress PE won't even know(need not know is better) if a packet is
> fragmented or not.
> 
> rgds,
> gopal
> 
> -----Original Message-----
> From: Ruyter Hill <Hill.Ruyter@carrier1.com>
> To: 'raszuk@cisco.com' <raszuk@cisco.com>; 'eosborne@cisco.com'
> <eosborne@cisco.com>
> Cc: MPLS OPS (E-mail) <mpls-ops@mplsrc.com>
> Date: Wednesday, August 15, 2001 9:14 AM
> Subject: Packet fragmentation on MPLS VPN
> 
> 
> >
> >Hi Robert, Eric
> >
> >I have an urgent question regarding MPLS VPN's I hope one of 
> you guys can
> >help me out ;o)
> >
> >
> >If I have a PE device receiving 4k packets on a Hssi that 
> need to be passed
> >across an MPLS vpn
> >as follows
> >
> >
> >( CE )---MTU 4k---( PE )---MTU 1500 ---( P )-------( P 
> )---MTU 1500---( PE
> >)---MTU 4k---( CE )
> >
> >when the packet leaves the (PE) for towards the (P) and the 
> packets need to
> >be fragmented will these be
> >re-integrated into a single packet at the opposite (PE) 
> before being passed
> >to the (CE) or will they be passed as the fragments
> >to the CE
> >
> >not sure as I don't know when fragmentation occurs as the 
> initial label is
> >added at the VRF and the second at the outgoing interface ( I think )
> >so in my understanding the packet would not be complete for 
> the VRF until
> it
> >was re-integrated at the opposite end
> >
> >If this is the case how will it effect PHP
> >
> >Kind Regards
> >
> >Hill Ruyter
> >
> >-------
> >The MPLS-OPS Mailing List
> >Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
> >Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
> 
> 
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
> 
> -------
> The MPLS-OPS Mailing List
> Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
> Archive: http://www.mplsrc.com/mpls-ops_archive.shtml
> 

-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml