I learnt that physical mtu should be as close to 1600 (1604)for expansion of services or higher , Protocol family mpls I would instinctively add an additional 26 bytes to 1552. to account for Q in Q and other possible expansion. The Juniper physical interface MTU is set at a value above the protocol MTU, in the case of Gigabit Ethernet the difference must be +26 to account for headers. the mpls MTU should be sufficient to meet the needs of the underlying routed and routing protocols. so IMHO to be generous with physical and MPLS might not be a bad thing, unless of course you encounter issues related to MTU handling in the hardware/OS deployed. I am sure there are arguments against as for. Would be pleased to hear them. Evan
Wisdom is what's left after we've run out of personal opinions. -Cullen Hightower ìÅÂÅÄÅ× éÇÏÒØ ÷ÌÁÄÉÍÉÒÏ×ÉÞ <[EMAIL PROTECTED]> wrote: Hello! Now we have cisco core network with MPLS MTU = 1526, for transit EoMPLS. 1518 - Ethernet + vlan .1q tag 8 - 2 MPLS Label 1518+8 = 1526 Example: interface GigabitEthernet0/1.15 description PE1 encapsulation dot1Q 15 ip address 1.1.1.30 255.255.255.252 tag-switching mtu 1526 tag-switching ip ! On Juniper box, MPLS mtu automatically recalculated from MTU physical interface. And now, to set up mpls mtu 1526 on Juniper box we have configuration: >show configuration interfaces ge-0/1/4 description "link to cisco CORE"; vlan-tagging; mtu 1556; unit 15 { description MSN-MFIST1-PE3; vlan-id 12; family inet { mtu 1500; address 1.1.1.29/30; } family mpls { mtu 1526; } } What correct physical MTU, MPLS MTU must setup on Juniper box for the organization of a joint CISCO and JUNIPER MPLS network. Igor. _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp