Browse Prior Art Database

Method to improve multicast path MTU discovery

IP.com Disclosure Number: IPCOM000031804D
Original Publication Date: 2004-Oct-11
Included in the Prior Art Database: 2004-Oct-11
Document File: 2 page(s) / 43K

Publishing Venue

IBM

Abstract

When a sender S sends a multicast packet to multicast group G with path MTU discovery turned on, it might get ICMP error messages back from a router indicating that an outgoing link has a smaller MTU than the current estimate. The sender S then lowers the path MTU estimate and starts sending out smaller datagrams. Consider the following network shown below: S is the sender of multicast datagrams. R1-R9 are multicast routers that forward multicast datagrams. In this case sender S initially sends 1500 byte multicast datagrams. However router R7 will send it an ICMP error message indicating that there is a smaller MTU of 1000 bytes. Sender S will then fragment the multicast datagrams and send them out. Notice that in this case, every other link in the multicast tree can support 1500 byte datagrams. They will all be penalized because of the one small link between R7 and R9.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 77% of the total text.

Page 1 of 2

Method to improve multicast path MTU discovery

A solution to this problem where sender S will discover the MTU sizes of the multicast tree and make a better decision about whether to fragment or not. In the above case, for example, it would be better if S continued to send 1500 byte datagrams which can be delivered as-is to every part of the network, except the R7-R9 section. In this case fragmentation will now only occur in the R7-R9 portion and the rest of the network will have lower network traffic.

R8

R9

1500

1500

1500

When S sends out the multicast datagram, it will include the IP address of the first hop multicast router R1 in this message as an IP option in the datagram.

Our solution will involve the DVMRP (Distance Vector Multicast Routing Protocol). This protocol is used by the multicast routers R1 through R9 to talk to each other and discover the multicast spanning tree for multicast group G. A new type of message will be added to the DVMRP protocol. When a multicast router sees the first datagram from

1000

R4

R5

R6

R7

1500

1500

1500

R2

R3

1500

R1

1500

S

1

[This page contains 4 pictures or other non-text objects]

Page 2 of 2

sender S to multicast group G, it must send a message to the first hop router R1 including:
1. Its own IP address
2. For each neighboring multicast router, the maximum transmission unit (MTU) that will be used to reach that router. (All the routers will know the IP address of the first hop router R1, since sender S has included it as...