System models 2 in distributed system

Tiered Architecture
• Technique to organize functionality of a given
layer and place this functionality into
appropriate servers.
• The functional decomposition is as
- Presentation Logic : Concerned with user
handling
- Application Logic: Detailed application –
specific processing
- Data Logic: persistent storage of application
Isha Padhy, Department of CSE 1

2 tier Architecture
• Server provides processing and data management; client provides simple graphical
display (thin-client)
• At the other extreme, all application processing and some data resides at the client
(fat-client approach)

Two- tier architecture
• Advantages:
1.Because of tight coupling the application runs
faster.
2.Low latency in terms of interaction.
• Disadvantage:
Invoking a part of process from other part
across the process boundary is difficult.
• Example- Railway reservation: User is using
railway reservation software, gives input and
then sends request to server. 3Isha Padhy, Department of CSE

3 Tier Architecture
An example of a server acting as client.
-In some applications servers may also need to be clients, leading to a three level
architecture. Ex. Web servers that interact with database servers
-Presentation logic: UI part of application, allows user to input data.
-Application Logic: All business logic is written like data insertion, validation etc
-Data Logic: Contains method to interact with database.

De-centralized Architecture
• Vertical distribution: Logically different components are
placed on different machines.
• Horizontal Distribution/ Peer to peer: All nodes are
independent and work on its share of data set.
- Why it was developed?
Network and computing resources owned by the users of a
service could also be utilized to support that service.
- Two types depending on how to organize the processes in a
n/w.
1. Structured
2. Unstructured
5Isha Padhy, Department of CSE

Structured
• N/w constructed using a procedure where
processes are organized using distributed hash
table(DHT).
• DHT provides a lookup service where (key,
value) pairs are stored. Data are assigned a
random key from identifier space, nodes are
assigned values from the same space.
• Responsibility for mapping from keys to
values is distributed among nodes.
• Ex Chord system 6Isha Padhy, Department of CSE

What is Chord?
• Steps in chord:
1. Find a value:
- Nodes are organized in a ring s.t. data item with key ‘k’ is mapped to node
with smallest identifier id~k. This node is called successor of key ‘k’,
Succ(k).
- A function is called to lookup for ‘k’ i.e LOOKUP(k) which returns n/w
address of succ(k)
2. Node joins a system: process starts with generating a random identifier
‘id’. Node that joins get ‘id’. Node does lookup(id) and gets n/w address of
succ(id). Then it inserts itself into ring.
3. Leaving: Node informs its departure to its predecessor and successor and
transfers its data to successor of node.

8
6
1
2
6
0
4
26
5
1
3
7
2
identifier
circle
identifier
node
X key
The Chord algorithm
successor(1) = 1
successor(2) = 3successor(6) = 0
Solves problem of locating a data item in a
collection of distributed nodes,
considering frequent node arrivals and
departures
Isha Padhy, Department of CSE

9
Node Joins and Departures
6
1
2
0
4
26
5
1
3
7
successor(6) = 7
6
1
successor(1) = 3
Isha Padhy, Department of CSE

Unstructured PTP
• Pure P2P: All nodes are of equal ability i.e no nodes have any
infrastructure, ex Gnutella n/w.
• Hybrid P2P: Infrastructure nodes are allowed to exist and are a
type of servers, all clients connected to the n/w must connect
to one of the servers. ex Napster n/w.
• Centralized P2P: Super-nodes are present in n/w which
dynamically service a small subpart of the peer n/w, ex Kazaa
n/w.
Requestor
node
3
2
1
4

Placement of objects and services
• Mapping of services to multiple servers: Servers may –
1. Partition of set of objects on which service is based
2. Replicate copies of them on several hosts
• Caching- Cache can be present in client side or proxy servers
that stores the recentlt used data objects
• Mobile code- applet code is stored in web server which can be
downloaded to a browser and run.
• Mobile agents- Composition of software and data that can
migrate from one system to other.

Hybrid Architectures
• Edge –Server Systems:
- Servers are placed “at the edge” of the network.
- Purpose is to serve content after filtering and transcoding
functions to data.

Hybrid Arch.
• Collaborative DS

Fundamental Model
• The purpose is
- All relevant assumptions about the DS can be made
- With the assumptions what are the possibilities
• Aspects to be considered are:
- Interaction: Considers the delay time taken by the
message from one process to other.
- Failure: Defines and classifies fault so design a system
to tolerate fault.
- Security: Analysis of threats to a system is done so that
system resists them.

Interaction Model
• Two factors affecting interaction of processes in DS
are:
- Performance of communication channels
- -Latency
- -Bandwidth
- -Jitter : Variation of time within two set of strings
transferred.
- Computer clocks and timing events : Two processes
running on different machines have different local
time and timestamps. So a global time standard is
used.
Time taken by first string of bits to reach
destination.
Congestion in network
Time taken by OS services in sending and
receiving side.

Interaction Model
• Two variants of IM
- Synchronous DS
--Strong assumption of time
--Bounds are defined for
1. Time to execute each step of a process which has known
lower and upper bounds
2. Each message transmitted over a channel is received within
a known bounded time
3. Each process has a local clock whose drift rate from real
time has a known bound.
- Asynchronous DS:
--Makes no assumption of time

Failure Model
• Defines the way in which failures may occur and
the effects of failures
• Types of failures
- Failures of processes
- Failures of communication channel
• Categories of failures
- Omission Failures
- Arbitrary Failures:
- Timing Failures
Process OF
Channel OF

Process has crashed i.e halted and will not
execute further.
Services execute properly if:
1.Processes either execute correctly or stop.
2.Follow time- outs i.e. a process allows a fixed
period of time for something to occur.
Process Omission Failures

Communication OF
Process p Process q
Communication channel
send
Outgoing message buffer Incoming message buffer
receivem
Failure occurs when a message is not transported from process ‘p’ outgoing
buffer to process ‘q’ incoming buffer due to lack of buffer space or n/w
transmission error.

Arbitrary Failures
• Arbitrarily omits intended processing steps or
takes unintended steps.
• Ex , A process may set wrong values in its data
items or may return a wrong value in
response.
• AF are uncommon in channels.

Timing Failures
• Synchronous DS:
- Failure of clock on process : Process local
clock exceeds the bound on its rate of drift
from real time.
- Failure of performance on process: Process
exceeds the interval between two steps
- Failure of performance on channel: A
message’s transmission takes longer time
• Asynchronous DS: Overloaded server may
respond late. 21Isha Padhy, Department of CSE

Security model
• Security model - Secure processes and channels
and protect objects encapsulated against unauthorized
access.
– Protecting access to objects - Access rights,
authentication of clients
– Protecting processes and interactions
– Protecting communication channel.

Security model
Network
invocation
result
Client
Server
Principal (user) Principal (server)
ObjectAccess rights
Communication channel
Copy of m
Process p Process qm
The enemy
m’
Protecting objects
Protecting process

Protecting channel
• Client and server knows the identity of each
other( Authentication Process)
• Secure channel ensures privacy and integrity
of data( cryptography process)
• Each message includes a logical timestamp to
prevent messages from being reordered.

System models 2 in distributed system

More Related Content

What's hot

Viewers also liked

Similar to System models 2 in distributed system

Recently uploaded

System models 2 in distributed system