Subject Name: DISTRIBUTED COMPUTING
Unit No: 1 Unit Name: Introduction to
distributed System Faculty Name:
Mrs.. HarshaSaxena
[Link] Alte
[Link]
Index
Lecture 1 – Introduction to distributed System:Definition,Goals,Issues 4
Lecture 2 – Introduction to distributed System:Types of distributed system 10
Lecture 3 –Introduction to distributed System:Distributed system models 16
Lecture 4 – Introduction to distributed System:Hardware and software concepts 22
Lecture 5-Introduction to distributed System:Models of middleware system,services
41
offered by middleware system
Lecture 6 –Introduction to distributed System:Client-server model 47
2
Unit No: 1 Unit name:Introduction to Distributed Systems
LectureNo:4
Hardware concept,Software
concept
Hardware Concepts
Distributed system consist of multiple CPUs and memory, there are several
different ways the hardware can be organized in terms of how they are
interconnected and how they communicate.
MULTIPROCESSOR MULTICO
MPUTER
4 Lecture 4-:Hardware concepts
Hardware Concepts Classification
• Multi Processor and Multi Computer
Multi Processor-Those have shared memory(Single Physical address space)
Multi Computer-has its own private memory
• Bus and Switched
Bus-Single network, backplane, cable that connect all the machine.
Switch-There are individual wires from machine to machine with many different
wiring pattern
• Homogenous and Heterogeneous (Distinction for only Multicomputer)
Homogenous-use same technology for interconnection, same access to private
memory( generally working on same problem)
Heterogeneous-Different independent computer connected through different
network through switch
5 Lecture 4-:Hardware concepts
Multiprocessors (1)
A bus-based multiprocessor
• Direct access to shared memory
• Connected to a common bus
• Memory should maintain coherent (Consistent)property
• With increase of CPU performance drop
• To increase performance, High-speed cache memory is used between CPU
and bus.
• It store recent data
• Can lead to incoherent memory and need replication mechanisms.
Lecture 4-:Hardware concepts
6
Multiprocessors (2)
A Switch-based multiprocessor
• for more than 256 CPU
• Memory is divided in modules
a) A crossbar switch-Each CPU and each
memory are connected.
• Switch closed when one CPU access the
memory. Rest CPU waits for switch to open.
• For n CPU and n memory,n^2 switches are
needed.(network is limited to less value of n)
b) An omega switching network
• In fig network contain 2X2 switch, each having
2 input and 2 output
• Disadvantage- There may be several switching
stages between the CPU
Lecture 4-:Hardware concepts
7
Homogeneous Multicomputer Systems
In CPU-CPU communication, the volume of traffic is of lower magnitude than
for CPU-Memory traffic.
• System Area Networks(SANs) –the nodes are mounted in a big rack and
connected through single, high-performance network.
• Bus-based Multicomputer(20-100 nodes)- Connected through shared
multi-access network such as Fast Ethernet.(message is broadcast )
• Switch-based Multicomputer- (message is routed)
a) Grid: 2-dimensional network
b) Hypercube: n-dimensional cube , each vertex is CPU
Lecture 4-:Hardware concepts
8
Software Concepts
An overview of
•DOS (Distributed Operating Systems)
•NOS (Network Operating Systems)
•Middleware
Lecture 22-:Software concepts
9
Computer Architecture
Computer architectures consisting of interconnected, multiple processors are
basically of two types:
• Tightly coupled systems
• Loosely coupled systems
10 Lecture 1-:Definition,Issues,Goals
Tightly Coupled System
• In these systems, there is a single system wide primary memory (address
space) that is shared by all the processors.
• If any processor writes, for example, the value 100 to the memory location x,
any other processor subsequently reading from location x will get the value
100.
• Therefore, in these systems, any communication between the processors
usually takes place through the shared memory.
100
11 Lecture 1-:Definition,Issues,Goals
Loosely coupled systems
• In these systems, the processors do not share memory, and each processor
has its own local memory.
• If a processor writes the value 100 to the memory location x, this write
operation will only change the contents of its local memory and will not affect
the contents of the memory.
• In these systems, all physical communication between the processors is
done by passing messages across the network that interconnects the
processors.
100
12 Lecture 1-:Definition,Issues,Goals
Software Concepts
System Description Main Goal
Distributed Tightly-coupled operating system for multi- Hide and manage
Operating processors and homogeneous hardware
System (DOS) multicomputers resources
Network Loosely-coupled operating system for Offer local
Operating heterogeneous multicomputers (LAN and services to remote
System (NOS) WAN) clients
Provide
Additional layer atop of NOS implementing
Middleware distribution
general-purpose services
transparency
Lecture 4-:Software concepts
13
1. Distributed Operating Systems (1)
• DOS functionality is same as traditional OS for uniprocessor system, except that
they handle multiple CPUs.
• OS should have full control on how the hardware resource are used and shared
1.11
Uniprocessor Operating System
14 Lecture 4-:Software concepts
1. Distributed Operating Systems (2)
• Access to Shared Memory
• High performance
• No of CPUs are transparent to application
Multiprocessor Operating System
15 Lecture 4-:Software concepts
1. Distributed Operating Systems (3)
General structure of a homogeneous multicomputers operating system
1.14
Multicomputer Operating Systems
Lecture 4-:Software concepts
16
1. Distributed Operating Systems (4)
Alternatives for blocking and buffering in message passing.
S1: Block sender when buffer is full
S2: Message being sent
S3: Message is arrived
S4: Message is sent to receiver
Lecture 4-:Software concepts
17
1. Distributed Operating Systems (5)
Relation between blocking, buffering, and reliable communications.
Reliable comm.
Synchronization point Send buffer
guaranteed?
Block sender until buffer not full Yes Not necessary
Block sender until message sent No Not necessary
Block sender until message received No Necessary
Block sender until message delivered No Necessary
Lecture 4-:Software concepts
18
1. Distributed Operating Systems (6)
• Programming a multicomputer is much harder than programming
multiprocessor.
• For communication to access shared data process in Multiprocessor used
semaphore and monitor for synchronization which is easier than message
passing.
• Buffering, blocking and reliable communication make things worse.
• To avoid this researcher provide virtual shared memory to multicomputer.
• Where user access large virtual address space which lead to page-based
distributed shared memory(DSM)
Lecture 4-:Software concepts
19
1. Distributed Operating Systems (7)
a) Pages (of size 4KB or 8KB) of address
space distributed among four machines
b) CPU1 Reference instruction 0,2,5,9, the
references are done locally
c) Situation after CPU 1 references page
10 is not present locally, a trap occurs
the OS fetch the page containing the
address and move it from CPU2 to
CPU1
d) This is normal paging concept, except
that remote RAM is used instead of local
disk.
e) Situation if page 10 is read only and
replication is used to increase
performance.
Distributed Shared Memory Systems
Lecture 4-:Software concepts
20
1. Distributed Operating Systems (8)
• Issue with DSM is to decide how large page should be.
• If page size is large, total number of transfer will be reduce which increase
performance
• A page contain two independent processes, OS need to consequently transfer the
page between two processor Known as False sharing
1.18
Lecture 4-:Software concepts
21
2. Network Operating System (1)
• Heterogeneous system
• Services Provided by NOS
• User can remotely login to another machine
• Transfer of file
• Global file server accessible to all
1-19
22 Lecture 4-:Software concepts
2. Network Operating System (2)
• Two clients and a server in a network operating system.
• File server maintain hierarchical file system each with a root directory containing
subdirectories and files.
1-20
Lecture 4-:Software concepts
23
2. Network Operating System (3)
Different clients may mount the servers in different places.
1.21
Lecture 4-:Software concepts
24
Difference between DOS and NOS
• DOS provide full transparency
• NOS lacks in maintaining transparency
• Disadvantage of NOS
• Its hard to login to every machine in network
• Cant use same password for every machine
• Protection against malicious attack
• Advantage of NOS
• Easy to add or remove machine/Server
• Easy to communicate existence of new machine to other
• Neither DOS or NOS qualifies as Distributed system definition
• We need Openness and Scalability of NOS
• And Transparency of DOS
Lecture 4-:Software concepts
25
3. Middleware(1)
Lecture 4-:Software concepts
26
Middleware
In an open middleware-based distributed system, the protocols used by each
middleware layer should be the same, as well as the interfaces they offer to
applications.
Lecture 22-:Software concepts
27
Services offered by Middleware System
There are many services offered by middleware system
[Link] service
[Link] service
[Link] service
[Link] service
[Link] service
[Link] service
Lecture 5-:Services offered by
28
Middleware system
Comparison between Systems
Distributed OS Network Middleware-
Item
Multiproc. Multicomp. OS based OS
Degree of
Very High High Low High
transparency
Same OS on all
Yes Yes No No
nodes
Number of copies of
1 N N N
OS
Basis for Shared Model
Messages Files
communication memory specific
Resource Global, Global,
Per node Per node
management central distributed
Scalability No Moderately Yes Varies
Openness Closed Closed Open Open
Lecture 4-:Software concepts
29
Thank You