Group Communication
Group oriented activities are steadily increasing.
There are many types of groups:
 Open and Closed groups
 Peer-to-peer and hierarchical groups
We focus on how the members of a group communicate
Major issues

Different types of multicast



Dynamic groups


Atomic multicast
Ordered multicast
How to communicate when the membership constantly changes
Failure handling

Keeping track of membership and membership changes
Atomic multicast
A multicast is called atomic, when the message is
delivered to every correct member, or to no
member at all.
How can we implement atomic multicast?
Basic vs. reliable multicast
Basic multicast does not consider crash failures.
Reliable multicast does.
Three criteria for basic multicast:
Liveness.
Each process must receive every message
Integrity.
No spurious message received
No duplicate. Accepts exactly one copy of a message
Reliable atomic multicast
Sender’s program
Receiver’s program
i:=0;
do i ≠ n 
send message to i;
i:= i+1
od
if m is new 
accept it;
multicast m;
 m is duplicate  discard m
fi
Tolerates process crashes.
Multicast support in networks
Sometimes, certain features available in the infrastructure
of a distributed system simplify the implementation of
multicast. Examples are


Multicast on an ethernet LAN
IP multicast
IP Multicast
Although multicasts can be implemented using point-to-point
communications, there are some practical forms of multicasts that
make use of the inherent multicasting ability of the underlying
medium. IP multicast is a bandwidth-conserving technology that
reduces traffic by simultaneously delivering a single stream of
information to multiple clients.
Applications that take advantage of multicast include distance
learning, videoconferencing, and distribution of software, stock
quotes, and news. The source sends only one copy, which is
replicated by the routers.
Distribution trees

Class D addresses are
assigned to multicast groups.
2
A
1
source
B
2
4

Routers maintain and update
distribution trees when
members join / leave a group.
D
Concern: Too much load on
routers. Application layer
multicast overcomes this.
F
15
C
E
6
7
(a) Source tree
rendezvous point
2

1
1
source
A
B
D
2
4
source
F
15
C
E
7
(b) Shared tree
1
6
Ordered multicasts
Total order
Causal order
Local order
(a.k.a. Single source FIFO)
Definitions?
Applications?

Total order multicast is useful
in the consistent update of
replicated servers

Causal order multicast is
relevant in implementing
bulletin boards

Local order multicast is useful
in updating cache memories
in multi-computers
Implementing total order multicast
Basic multicast using a
sequencer
{The sequencer S}
define seq: integer (initially 0}
do receive m 
multicast (m, seq);
seq := seq+1;
deliver m
od
sequencer
Implementing basic total order
multicast
Distributed implementation without a sequencer
Uses the idea of 2PC
p
q
r
3
18
4
22
6
19
7
10
14
Implementing basic total order
multicast




Step 1. Sender i sends (m, ts) to all
Step 2. Receiver j saves it in a holdback queue, and
sends an ack (a, ts)
Step 3. Receive all acks, and pick the largest ts. Then
send (m, ts, commit) to all.
Step 4. Receiver removes it from the holdback queue
and delivers m in the ascending order of timestamps.
Why does it work?
Implementing basic causal order
multicast
Use vector clocks.The
recipient i delivers a
message from j iff
1,0,0
0,0,0
(1,1,0)
P0
m1
1. VCj(j) = LCj(i) +1
{LC = local vector clock}
m1
m2
m3
1,1,0
0,0,0
P1
(2,1,0)
(1,0,0)
m2
2. k: k≠j :: VCk(j) ≤ LCk(i)
m3
2,1,1
0,0,0
VC = incoming vector clock
LC = Local vector clock
2,1,0
P2
(1,0,0)
(1,1,0) (2,1,0)
? (violation)
Note the slight difference in the
implementation of the vector clocks
Reliable multicast
Tolerates process crashes. The additional requirements are
as follows:
Only correct processes are required to receive the
messages from all correct processes in the group.
Multicasts by faulty processes will either be received by
every correct process, or by none at all.
A theorem on reliable multicast
In an asynchronous distributed system, total order reliable
multicasts cannot be implemented when even a single
process undergoes a crash failure.
Why? Since it will violate the FLP impossibility result.
Scalable Reliable Multicast
IP multicast or application layer multicast has to
detect the loss of messages and use
retransmission for achieving reliability. For large
groups (like distance learning applications)
scalability is a major problem.
Scalable Reliable Multicast

Difficult to scale:
Sender state explosion
 Message implosion

State:
receiver 1,
receiver 2,
…
receiver n
Scalable Reliable Multicast

If omission failures are rare, then receivers will only report the
non-receipt of messages using NACK, This has the potential to
trigger selective point-to-point retransmission The reduction of
acknowledgements is the underlying principle of Scalable
Reliable Multicasts (SRM).

If several members of a group fail to receive a message, then
each such member waits for a random period of time before
sending its NACK. This helps to suppress redundant NACKs.
Sender multicasts the missing copy only once.