[Search] [txt|pdf|bibtex] [Tracker] [Email] [Diff1] [Diff2] [Nits]

Versions: 00 01                                                         

Internet Drat                                             Man Yeob Lim
<draft-lim-ip-reliable-multicast-01.txt>                 Dae Young Kim
                                                   Chungnam Nat. Univ.
                                                              May 1998

                     IP Extension for Reliable Multicasting

Status of Memo

This document is an Internet Draft. Internet Drafts are working
documents of the Internet Engineering Task Force(IETF), its areas, and
its working groups. Note that other groups may also distribute working
documents as Internet-Drafts.

Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference material
or to cite them other than as "work in progress".

To view the entire list of current Internet-Drafts, please check
the "1id-abstracts.txt" listing contained in the Internet-Drafts
Shadow Directories on ftp.is.co.za (Africa), ftp.nordu.net
(Northern Europe), ftp.nis.garr.it (Southern Europe), munnari.oz.au
(Pacific Rim), ftp.ietf.org (US East Coast), or ftp.isi.edu
(US West Coast).


This memo presents IP extension for recovering multicast packets from
congestion. Dropped packets can be recovered far faster by IP routers
with extension of this memo than by group member end-hosts. Because
necessary interactions are limited among adjacent routers, this scheme
substantially reduces overall signaling overhead among group members
for packet recovery.

Lim & Kim                Expires November 1998                [page 1]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

1. Introduction

Since the IP Multicast was proposed [1], there have been many research
works on reliable multicast protocols. However the fact that the
multicast itself is done in the IP layer but the solutions are sought
in the transport or higher layer makes the search for solutions more
difficult. The transport protocol sits on group members' end hosts
which are spread over a large geographical area, and so if packet
losses occur in the network, it not only takes long to detect in the
transport layer but also makes group coordination very complicated.
Even though many schemes were proposed to overcome the multicast
communication losses[2-6], it is hard to devise a general solution
without any attempt involving the IP protocol in the task.

There are two types of packet losses in Internet environments. One type
is transparent to routers, while the other is not. The first type
includes packet losses due to link error or router failure. Because
these losses occur outside of the routers, in order to recover these
packets an end-to-end ACK/NAK operation is required. The second type
includes packet drops due to congestion. This type of packet loss is
made by explicit router decision when the router encounters congestion.
As the transmission quality improves, the first type of packet loss is
diminishing and the congestion becomes major reason for packet loss.
Because packet drops at congestion are done with routers' knowledge, we
can think of a recovery scheme by explicit coordination among routers.
If recovery of lost packets is done instantaneously and actively by the
IP routers before later intervention by the higher protocol, not only
the end-to-end multicast protocol can be significantly simplified but
also the recovery can be done in a much faster fashion. A minimal
requisite for the routers' capability at congestion in order to make
the proposed scheme possible is that the router should be able to see
the packet to collect necessary information before actually dropping

A study [7] shows that the loss on the links of the multicast network
is observed to be only 2% or less of the whole packet loss and also
that the rest congestion loss are again classified into two types,
single and burst.  Most of the congestion losses consist of isolated
single losses, but a few of very long loss bursts, lasting from a few
seconds up to 3 minutes(around 2000 consecutive packets) contribute
heavily to the total packet loss. If retransmission request is made
before congestion is relieved, the traffic will further become worse
and congestion will be extended. The appropriate congestion control
mechanism is required to cope with the burst loss.

The multicast packets are further categorized into two groups; real
time packets and non real time packets. The real time multimedia
packets are time sensitive packets like audio and video that should be

Lim & Kim                Expires November 1998                [page 2]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

recovered within a very short time if they are to be meaningful. These
packets also have the best-effort characteristics, that is, the more
packets are transferred the better quality is provided. Most upper
layer protocols take long time to recover any loss depending on the
location where the congestion occurs. If the congestion occurred
upstream, the distance between the unreceived members and the received
members are far apart. Even though the congestion occurred close to the
receivers it takes time for the group members to coordinate to recover.
Recovery by coordination among group members is not likely to meet real
time requirement. Instead, the feasible way is to recover by
coordination between adjacent routers.

The non real time multicast packets do not require critical timing
requirement but typically are very critical to multicast coordination
and reliability should be completely guaranteed. If these information
is lost then the conference coordination collapses and too much
endeavor to restore the control is required with time delay.

We propose extensions to IP and ICMP protocols for efficient recovery
of both real time and non real time packets dropped from router
congestion. We give multicast routers recovery cache or buffer so that
lost packets can be recovered by coordination among routers. It is not
suitable that all lost packets be recovered by routers. Recovery should
be limited only to important multicast packets which are specially
tagged, so that the cache/buffer size can be minimized and multicast
routers are not required to do too much a processing overhead. In IP
version 4, reliability bit in the type of service field can be used to
identify multicast packets which require recovery. Low delay bit in the
type of service can specify fast recovery using cache routers. In IP
version 6, one entry in the priority field is suitable to specify
reliable multicasting per packet or flow label can be used to specify
reliable multicasting or fast recovery per data stream.

2. Recovery Operation
2.1. Overview

When a router receives more data than it can handle, congestion occurs.
Once a congestion occurs the router receives no more data until the
congestion is removed. Because the input queue is full, the incoming
packets are not accepted in the queue but just ignored. We propose to
allocate small size of extra queue to store header of dropped packets.
The IP header is 20 bytes long and this information is enough to search
the same packet from duplicated packet storage. The header includes
source address, destination group address, identity and offset fields.
The identity field identifies packets out of a packet stream which is
going from one source to one destination. This field was defined for
packet segmentation and reassembly. When a packet is travelling through

Lim & Kim                Expires November 1998                [page 3]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

a network the packet is segmented into pieces if the underlying network
does not support long packet size. The segmented packets are
reassembled in the destination host based on the identity and the
offset field. If copies of the dropped packets are stored in a place it
is possible to identify the same packet with the identity and the
offset field. In order to retransmit the dropped packet, the congested
router requests retransmission sending a request packet which includes
header of the dropped packet.

Retransmission can be made by the source host whose IP address is in
the header. As packets are dropped farther from the source host, it
takes longer to recover by the sending host. We propose multicast
routers have internal buffer to hold duplicate copies of the multicast
packets as long as the packets can reach the next multicast routers.
The multicast routers are not all equipped with buffer, but routers in
every several hops in the routing tree are selected to have buffer. As
a packet travels along the routing tree, it passes buffering routers
once in several. If congestion occurs and a packet is lost, the packet
is repaired by the buffering router which the packet passed at last.
For this purpose we add an option field to the IP data packet to store
the last buffering router's IP address. The address field is updated by
every buffering router when the packet passes through the router.

If the buffering router receives repair request, the router searches
the duplicate packet in the buffer and encapsulates the packet into a
unicast one. The unicast packet is transferred to the requesting router.
Then the requesting router restores multicast packet and resume

When a host is located close to a multicast router and is available as
a buffering device instead of the buffering router itself, the host can
be used to store copies of multicast packets. If a host registers
itself as a buffering device to a multicast router, the router sends
all duplicated multicast packets  which pass through the router to the
buffering host and updates the buffering router address in the option
field with the IP address of the buffering host.

2.2. Delay before retransmission

When congestion occurs there arises a question that how soon the router
will be recovered from congestion and become ready to receive packets.
If the congestion extends for a long period of time fast retransmission
is useless or makes problem even worse. In reliable unicast protocol,
congestion control is provision of TCP, which seems working acceptable
because there are enough delay before the congestion is detected by the
end host even though the response is not optimally fast. In multicast
congestion makes reliable transport protocol harder and getting an
optimum congestion control even worse. The optimum scenario against

Lim & Kim                Expires November 1998                [page 4]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

congestion is to detect a congestion as soon as possible, to reduce
transmission rate and to wait until the congestion is removed before
any retransmission is attempted. This IP extension detects congestion
at the same time as congestion occurs because recovery operation is
initiated by the router itself which is congested. We propose to make
repair request after the congestion is removed. Because the header size
is only 20 bytes long there is no big impact to the memory usage even
though thousand headers are saved before requesting requests.

2.3. Recovering burst loss

When a burst loss occurs and thousand packets are lost, thousand repair
requests should be sent. In order to reduce the number of repair
request packets we can combine the requests into one or more combined
repair request. Because the routers are several hops in between, and
the routing table is not updated in small amount of time the IP packets
are travelling in order as they were produced, i.e. in sequence same as
the identity number. When congested router requests a repair, it can
analyze the headers and combine as many headers in series into one,
having same source and destination addresses. The repair request can be
to transmit packets from  source S to destination D with identity
between M and N. Even though there is one or several holes in the burst,
the retransmission can include them only to be ignored in the
destination host. Using this expression the requesting router can
reduce the memory holding headers of the lost burst packets.

2.4. Recovering real time packets

Real time packets are recovered by the so called cache router, which is
located at just one previous hop from where congestion occurs. Cache
routers continuously copy all multicast packets with service type of
low delay in its ring type cache. When a cache router forwards a
multicast packet, it updates the cache router address in the option.
While this multicast packet travels along the network toward
destination, the cache router address is updated every time the packet
passes through cache routers. When a router has to drop a packet due to
congestion, it sends a repair request to the cache router whose IP
address is specified in the option field of the packet. Upon reception
of the request, the cache router looks for the same packet in the cache.
If it is still there, the cache router retransmits the packet. If
repair request is made delayed the packet may not be recovered because
the cache is overwritten with new incoming packets. Knowing that the
real time packets are recovered by the just previous routers from
congestion point, all routers should be equipped with cache in order
for a network provides reliable multicast on real time packets.

2.5. Congestion control

Lim & Kim                Expires November 1998                [page 5]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

When a congestion occurs, there need a mechanism to control congestion.
Even though recovery mechanism against burst loss is prepared, if data
rate from source is not suppressed the congestion lasts longer and the
recovery operation might finally be overwhelmed. Because congestion is
both detected and handled in the IP layer congestion control should be
incorporated by IP protocol also. When a router detects internal queue
usage reaches close to the congestion state, for example 90%, the
router sends an Explicit Congestion Notice (ECN) to the source host, so
that the source host reduces data rate. Because it takes certain amount
of time for ECN travels to the source host and reduced packet stream
reaches the congestion point, the pending packets may cause congestion.
In order to reduce this possibility the ECN is sent earlier reserving
larger queue free.

We propose additional congestion control scheme which distributes the
congestion status to adjacent routers and share multiple queues against
congestion. If adjacent routers are notified that congestion occurred
on the next router, then the adjacent routers queue the packets which
are to go to the congested router as much as the queue size allows. If
the queue becomes full and another congestion happens or queue usage is
over the predetermined threshold then congestion is notified to the
adjacent routers again thus the congestion statues is propagated
outward from the congestion point. If the congestion at the initial
congestion point is relieved then this is notified to the adjacent
routers and packet forwarding is resumed. This congestion control
scheme can distribute congestion at specific router to many routers in
wide area dynamically using queues of those routers combined
efficiently. Disadvantage is that congestion may happen in many routers
simultaneously, thus overall network goes down instead one single
router. Instead of using this scheme separately, using together with
the recovery operation and congestion control of source host may
accomplish better performance.

3. Protocol Data Unit(PDU) description
3.1. Extension to IP datagram adding cache/buffering router address

An option is defined in IP datagram to store cache or buffering router
IP address. Figure 1 shows packet format of the option in the IP

 0               8              16              24            31
 |      code     |    length     |           reserved          |
 |                       Router IP Address                     |

 Figure 1.  The format of the router option in an IP datagram

Lim & Kim                Expires November 1998                [page 6]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

The source host initializes the option field as the IP address of the
source host. Cache routers or buffering routers updates the field as
its IP address, selectively based upon the packet type which is
identified by the low delay priority bit; real time packet vs. non real
time packet. If a host is used as a buffering device, this field is
updated as the IP address of the buffering host. This makes it possible
that recovery is implemented by the nearest router from the congested

3.2. Repair request ICMP Message

This message is sent to the cache/buffer router or buffering host when
a drop from congestion occurs. Receiving the request the router
searches the requested packet in the cache/buffer. The router sends the
packet to the congested router. The buffering router converts the
multicast packet into a unicast packet and sends to the congested
router. One message can request single or multiple packets. The
no_of_identity field specifies the number of packets with different
identity number. Because there are multiple packets with same identity,
this number is not same as the number of packets.   Figure 2 shows the
format of the drop ICMP message.

 0               8              16              24            31
 |      type     |      code     |           checksum          |
 |        no_of_identity         |           reserved          |
 |                        internet header                      |

 Figure 2. Repair request ICMP message format

3.3. Repair packet

When a buffering router receives a repair request then the router
searches the packet in the buffer. If it succeeds finding the packet it
converts the packet into a unicast packet saving the multicast address
in the option field and changing the destination address to the
congested router address. This repair packet is tunneling routers which
received original packet without congestion so that packet is recovered
where congestion occurred and no duplicate retransmissions happen.
Receiving router should convert the unicast packet to a multicast
packet and continue multicast routing. The figure 3 shows the format of
the multicast address option. Retransmission packet by cache routers is
forwarded in multicast format because the cache router is located
adjacent to the congested router.

Lim & Kim                Expires November 1998                [page 7]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

 0               8              16              24            31
 |     code      |     length    |           reserved          |
 |                       multicast address                     |

 Figure 3. The format of the multicast address option in repair packet

3.4. Register buffering host ICMP message

This message is used for a host to register itself to a multicast
router as a buffering host. When a multicast router receives this
message the router forwards all multicast packets with reliability
service type to the buffering host. Upon receiving repair request from
a congested router the buffering host transmits to the congested router
in unicast format.

 0               8              16              24            31
 |     type      |      code     |           checksum          |
 |                   buffering host IP address                 |

 Figure 4. Register buffering host ICMP message format

3.5. Flow control ICMP message

This message is used to control flow rate from source. The flow control
figure means the router state how fast the router can process packets.
The figure is expressed from 0 to 255. Figure 0 means completely free
and figure 255 means that router is completely blocked.

 0               8              16              24            31
 |     type      |      code     |           checksum          |
 |  flow control |                    reserved                 |

 Figure 5. Flow control ICMP message format

3.6. Congestion propagation ICMP message

This message is used to notify adjacent routers that congestion
occurred so that forwarding is suppressed. This message is broadcasted
with TTL set to 1, i.e. only the surrounding routers can receive this

Lim & Kim                Expires November 1998                [page 8]


Internet Drat     IP Extension for Reliable Multicasting      May 1998

packet. The same message is also used when the congestion is removed
with different code number. Therefore all routers have information
whether the adjacent routers are working or blocked. If one router is
blocked no packet is forwarded to the router. Instead the packets are
buffered until it becomes unblocked. If this causes the router itself
becomes congested then congestion is again propagated to the next
surrounding routers and there is no more packets incoming.

 0               8              16              24            31
 |      type     |      code     |           checksum          |

 code c1: congestion occurred
 code c2: congestion removed

 Figure 6. Congestion propagation ICMP message format

4. Implementation issues

Suppose cache routers are used in a gigabit network and routers are
separated by 100 kilometers apart. Packet travel time is 0.3
millisecond for one way. If a congestion occurs, the cache router drops
a received multicast packet and sends a repair request. If we assume
the time to process a received packet and to generate a request is 0.4
millisecond, the cache router should store a duplicate copy of a
multicast packet by 1 millisecond. This results in 1 Mbit cache
required for each channel.

Considering buffer size of a buffering router, suppose round travel
time between source and destination host is 10 seconds. If we give
router's recovery time to recover from congestion 10 seconds the total
time to store packets in the buffer will be 20 seconds. Supposing 1
percent of the total 1 gigabit traffic is multicasting traffic
requiring retransmission, the buffer size will be 25 Mbyte. If we
increase the recovery time to 3 minutes the buffer size becomes around
250 Mbyte, and we feel this figure is not difficult to implement.

Lim & Kim                Expires November 1998                [page 9]


Internet Drat     IP Extension for Reliable Multicasting      May 1998


Man Yeob Lim, Mr                        Dae Young Kim, Prof.
InfoCom Eng. Dept.                      InfoCom Eng. Dept.
Chungnam National University            Chungnam National University
Daejeon 305-764                         Daejeon 305-764
Korea                                   Korea

Phone: +82 42 821 3544                  Phone: +82 42 821 6862
Fax:   +82 42 821 2225                  Fax:   +82 42 823 5586
Email: mylim@sunam.kreonet.re.kr        Email: dykim@ccl.chungnam.ac.kr


[1]  S. Deering, Host Extensions for IP Multicasting, RFC 1112, Jan.
[2]  S. Kasera, J. Kurose, and D. Towsley, Scalable reliable multicast
using multiple multicast groups, Proc. ACM Sigmetrics Conference,
[3]  S.Floyd, V. Jacobson, C. Liu, S. McCanne, and L. Zhang, A
reliable multicast framework for light-weight sessions and
application level framing, ACM SIGCOMM 95.
[4]  S. Armstrong, A. Freier, K. Marzullo, Multicast Transport
Protocol, RFC 1301, Feb. 1992.
[5]  B. Whetten, T. Montgomery, S. Kaplan, A high performance totally
ordered multicast protocol, Theory and Practice in Distributed
Systems, Springer Verlag, LCNS 938.
[6]  C. Papadopoulos, G. Paruklar, G. Varghese, An error control
scheme for large-scale multicast applications, Washington
University, St. Louis.
[7]  M. Yajnik, J. Kurose, and D. Towsley, Packet loss correlation in
the Mbone multicast network, University of Massachusetts at

Lim & Kim                Expires November 1998               [page 10]