DCS Unit-II COMMUNICATION AND COORDINATION.pdf

 Introduction
 Layered protocol
 Types of communication
 Network virtualization
 Network coordination

It is the communication amongst several
processes involved within a distributed system
scenario.
A process is a program in execution.
Process
1
Process
2
Data
Figure.1) Interprocess communication in distributed
system

Process 1
Process 2
Process 1
Process 2 Process 3 Process n
M
M
M
M
---
Fig.1.a) Unicast IPC Fig.1.b) Multicast IPC

I. Synchronous and asynchronous
communication
 In the synchronous form, both send and
receive are blocking operations.
 In the asynchronous form, the use of the send
operation is non-blocking and the receive
operation can have blocking and non-blocking
variants.

II. Message destinations
 A local port is a message destination within a
computer, specified as an integer.
 A port has an exactly one receiver but can have
many senders.
III. Reliability
 A reliable communication is defined in terms of
validity and integrity.
 A point-to-point message service is described as
reliable if messages are guaranteed to be
delivered despite a reasonable number of packets
being dropped or lost.
 For integrity, messages must arrive uncorrupted
and without duplication

IV. Ordering
 Some applications require that messages be
delivered in sender order.
V. Persistent and Transient Communication
 An electronic mail system is a classic example
where communication is persistent.

The java API for inter-process communication in the
internet provides both datagram and stream
communication.
The two communication patterns that are most commonly
used in distributed programs:
I. Client-Server Communication.
II. Group Communication.
Protocols involved in interprocess communication are as
follows:
I. TCP(Transport Control Protocol).
II. UDP(User Datagram Protocol).

Message Passing:
The application program interface (API) to UDP
provides a message passing abstraction.
Message passing is the simplest form of inter-
process communication.
API enables a sending process to transmit a
single message to a receiving process.
The independent packets containing theses
messages are called datagrams.
In the Java and UNIX APIs, the sender
specifies the destination using a socket.

 Request-reply protocols are designed to support
client-server communication in the form of either
RMI or RPC.
 Group multicast protocols are designed to
support group communication.

 The request and reply messages provide the basis for
remote method invocation (RMI) or remote procedure call
(RPC).
 Remote Method Invocation (RMI)
 It allows an object to invoke a method
in an object in a remote process.
 E.g. CORBA and Java RMI
 Remote Procedure Call (RPC)
 It allows a client to call a procedure in
a remote server.

Figure 3. Request-reply communication

The information to be transmitted in a request
message or a reply message is shown in below.
Figure 4. Request-reply message structure

invocation invocation
remote
invocation
remote
local
local
local
invocation
invocation
A
B
C
D
E
F
Figure 4. Remote and local method
invocations

 Characteristics:
 End point for inter-process communication.
 Message transmission between sockets.
 A socket is associated with either UDP or TCP. Sockets
are bound to ports.
 One process can use many ports .
 Processes don’t share sockets (unless for IP multicast).
 ƒ
Implementations :
 originally BSD Unix, but available in Linux, Windows,…
 APIs in programming languages (e.g., java.net)

 Send
 send a message to a socket associated to a
process .
 can be blocking or non-blocking.
 ƒ
Receive
 receive a message on a socket .
 can be blocking or non.
 Blocking.
 ƒ
Broadcast/Multicast
 send to all processes/all processes in a group.

Figure.5) Communication Using Socket

Marshalling and Unmarshalling
 ƒ
Marshalling: Encode data items so that they
can be written onto a stream.
 ƒ
Unmarshalling: Read an encoding from a
stream and reconstruct the original items.
Examples:
 ƒ
CORBA: CDR(= Common Data
Representation) for primitive and structured
data types that occur in remote method
invocations.
 ƒ
Java Serialization: Applicable to all classes
that implement the interface Serializable.

 The client-server communication is designed to support
the roles and message exchanges in typical client-server
interactions.
 In the normal case, request-reply communication is
synchronous because the client process blocks until the
reply arrives from the server.
 Asynchronous request-reply communication is an
alternative that is useful where clients can afford to
retrieve replies later.
 Avoidance of connection establishment overhead
 No need for flow control due to small amounts of data
are transferred.

The pair wise exchange of messages is not the best
model for communication from one process to a
group of other processes.
A multicast operation is more appropriate.
Multicast operation is an operation that sends a
single message from one process to each of the
members of a group of processes.
Multicasting has the following characteristics:
 Fault tolerance based on replicated services
o A replicated service consists of a group of
servers.

IP Multicast
 IP multicast is built on top of the Internet
protocol, IP.
 IP multicast allows the sender to transmit a
single IP packet to a multicast group.
 A multicast group is specified by class D IP
address for which first 4 bits are 1110 in IPv4.
The membership of a multicast group is dynamic.
A computer belongs to a multicast group if one or
more processes have sockets that belong to the
multicast group.

The following details are specific to IPv4:
 Multicast IP routers
 IP packets can be multicast both on local network and on
the wider Internet.
 Local multicast uses local network such as Ethernet.
 To limit the distance of propagation of a multicast
datagram, the sender can specify the number of routers it
is allowed to pass- called the time to live, or TTL for short.
 Multicast address allocation
 Multicast addressing may be permanent or temporary.
 Permanent groups exist even when there are no members.
 Multicast addressing by temporary groups must be created
before use and cease to exit when all members have left.

Java API to IP Multicast:
 The Java API provides a datagram interface to IP
multicast through the class Multicast Socket, which is a
subset of Datagram Socket with the additional
capability of being able to join multicast groups.
 The class Multicast Socket provides two alternative
constructors , allowing socket to be creative to use
either a specified local port, or any free local port.

Multicast Transmission
A message sent to a specified group of recipients.
Examples
 ƒ
Fault tolerance: Based on replicated services –
requests go to all servers
 ƒ
Spontaneous networking: All nodes of the network
receive messages
 ƒ
Better performance: Through replicated data –the
updated data goes to all storing the data.
 ƒ
Event notification

Types Of Overlay Networks:
1. Distributed hash tables.
2. Peer-to-peer file sharing.
3. Content distributed networks
4. Wireless ad hoc networks.
5. Disruption-tolerant networks.
6. Multicast.
7. Resilience.
8. Security.

Process 1 Process 2
Figure.6)Simple message passing communication
paradigm

1. Simplicity
2. Uniform semantics
3. Efficiency
4. Reliability
5. Correctness
6. Flexibility
7. Security
8. Portability

Send m
……………….
Receive
……………….
message
Figure.7)The architectural model of MPI
Process P Process q

A. Generic send operation
B. Synchronous send operation
C. Buffered send operation
D. Ready send operation

Figure 7. Request-reply communication
Request-Reply Communication

Figure 8. Request-reply message structure

Categories of Request-Reply protocols:
A. The Request Protocol
B. The Request/Reply protocol
C. The Request/Reply/Acknowledge-
Reply Protocol

Figure 9. Open and closed groups

Figure 10. Group membership management

Figure 12.a) Architecture of Public-Subscribe

Figure 12.b) Public-Subscribe Paradigm

Figure 13. Message Queue Paradigm

Figure 14. Topology In WebSPHERE

Measuring Time :
 Traditionally time measured astronomically .
1. Transit of the sun (highest point in the sky)
2. Solar day and solar second
 Problem: Earth’s rotation is slowing down.
1. Days get longer and longer
2. 300 million years ago there were 400 days in the year
 Modern way to measure time is atomic clock.
1. Based on transitions in Cesium-133 atom
2. Still need to correct for Earth’s rotation
 Result: Universal Coordinated Time (UTC).
1. UTC available via radio signal, telephone line, satellite (GPS

 External synchronization
 Internal synchronization

 Cristian’s Algorithm
 Berkeley Algorithm
 Network Time Protocol

Figure 14. Cristian’s Algorithm

Current time from a time server: UTC from
radio/satellite etc Problems:
 time must never run backward
 variable delays in message passing
Calculation:
 Earliest time that S can have sent reply: t0 +
min
 Latest time that S can have sent reply: t0 + T
round – min
 Total time range for answer: T round - 2 *
min
 Accuracy is ± (½Tround - min

Berkeley algorithm
 No external synchronization, but one master
server
 Master polls slaves periodically about their
clock readings
 Estimate of local clock times using round
trip estimation
 Averages the values obtained from a group
of processes - Cancels out individual clock’s
tendencies to run fast
 Tells slave processes by which amount of
time to adjust local clock
 Master failure: Master election algorithm

Goals
1. Ability to externally synchronize clients via
Internet to UTC
2. Provide reliable service tolerating lengthy
losses of connectivity
3. Enable clients to resynchronize sufficiently
frequently to offset typical HW drift rates
4. Provide protection against interference

NTP Basic Idea
 Layered client-server architecture, based on
UDP message passing
 Synchronization at clients with higher strata
number less accurate due to increased
latency to strata 1 time server
 Failure robustness: if a strata 1 server fails, it
may become a strata 2 server that is being
synchronized though another strata 1 serve

Procedure-Call and Symmetric Modes
 All messages carry timing history information.
 Local timestamps of send and receive of the previous
NTP message.
 Local timestamp of send of this message.
Figure 16. Procedure-Call and
Symmetric Modes

NTP: Delay and Offset
Figure 17. NTP: Delay and Offset

NTP Implementation
 Statistical algorithms based on 8 most recent <oi, di>
pairs: à determine quality of estimates
 The value of oi that corresponds to the minimum di is
chosen as an estimate for o
 Time server communicates with multiple peers, eliminates
peers with unreliable data, favors peers with higher strata
number (e.g., for primary synchronization partner
selection)
 NTP phase lock loop model: modify local clock in
accordance with observed drift rate
 Experiments achieve synchronization accuracies of 10
msecs over Internet, and 1 msec on LAN using NTP

Example: Totally-Ordered Multicasting(1)
Figure 18. Totally-Ordered Multicasting
(1)

Example: Totally-Ordered Multicasting (2)
Figure 18. Totally-Ordered Multicasting
(2)

Figure 19. Ad-hoc State Snaphots

A logical ring constructed in software.

 Group {Pi} ”fully connected”; election: ring
 Pi notices: coordinator lost Send
ELECTION(Pi) to the next P
 Pj receives ELECTION(Pi) Send ELECTION(Pi,
Pj) to successor
 n. . .
 Pi receives ELECTION(..., Pi, ...) active_list =
{collect from the message}
NC = max {active_list}
Send COORDINATOR(NC; active_list) to the
next P
A Ring Algorithm (2)
A Ring Algorithm (2)

DCS Unit-II COMMUNICATION AND COORDINATION.pdf

More Related Content

Similar to DCS Unit-II COMMUNICATION AND COORDINATION.pdf (20)

More from rameshwarchintamani (9)

Recently uploaded (20)

DCS Unit-II COMMUNICATION AND COORDINATION.pdf