Introduction to

Distributed Systems

INF5040/9040 Autumn 2018

Lecturer: Eli Gjørven (ifi/UiO)

August 28, 2018

 Outline

▪ Definition of a distributed system

▪ Goals of a distributed system
▪ Implications of distributed systems
▪ Pitfalls in developing distributed systems
▪ Types of distributed systems

INF5040, ifi/UiO 2
 From a Single Computer to DS

1945-1985: Computers
 Large and expensive
 Operated independently
1985-now: two advances in technology
 Powerful microprocessors
 High-speed computer networks
Result: Distributed Systems

Putting together computing systems composed of a large

number of computers connected by a high-speed network

INF5040, ifi/UiO 3
 What Is a Distributed System?

Definition
Operational perspective:
A distributed system is one in which hardware or software
components, located at networked computers, communicate
and coordinate their actions only by passing messages.
[Coulouris]

User perspective:
A distributed system is a collection of independent computers that
appears to its users as a single coherent system.
[Tanenbaum]

INF5040, ifi/UiO 4
 What Is a Distributed System?

A DS organized as middleware
 extending over multiple machines
 offering each application the same interface

INF5040, ifi/UiO 5
 Examples of Distributed Systems

Web search
 Indexing the entire contents of the Web
Massively multiplayer online games
 Very large number of users sharing a virtual world.
Financial trading
 Real time access and processing of a wide rage of
information sources.
 Delivery of items of interest in a timely manner

INF5040, ifi/UiO 6
 Outline

▪ Definition of a distributed system

▪ Goals of a distributed system
▪ Implications of distributed systems
▪ Pitfalls in developing distributed systems
▪ Types of distributed systems

INF5040, ifi/UiO 7
 Goals of Distributed Systems

Resource sharing
Distribution transparency
Openness
Scalability
Fault tolerance
Allowing heterogeneity

INF5040, ifi/UiO 8
 Resource Sharing

Making resources accessible:

 accessing remote resources
 sharing them in a controlled and efficient way
Examples: printers, storages, files, etc.
One reason to share: economics
 e.g., 1-to-many printer access, rather than 1-to-1
Resource managers control access, offer a
scheme for naming, and control concurrency
A resource sharing model describes how
 resources are made available
 resources can be used
 service provider and user interact with each other

INF5040, ifi/UiO 9
 Models for Resource Sharing

 Client-server resource model

 Server processes act as resource managers, and offer services
(collection of procedures)
 Client processes send requests to servers
 HTTP implements a client-server resource model
 Object-based resource model
 Any entity in a process is modeled as an object with a message
based interface that provides access to its operations
 Any shared resource is modeled as an object
 Object-based middleware (CORBA, Java RMI) defines object-based
resource models

INF5040, ifi/UiO 10
 Distribution Transparency

An important goal of a DS:

 hiding the fact that its processes and resources are
physically distributed across multiple computers
Definition:

A distributed system that is able to present itself to

its users and applications as if it were only a single
computer system is said to be transparent.

What kind of transparency?

INF5040, ifi/UiO 11
 Forms of Transparency

 Degree of transparency
 Situations in which full transparency is not good – better to
expose than to mask effects of distribution?
 Trade-off between a high degree of transparency and
performance

INF5040, ifi/UiO 12
 Openness

 Definition:

An open distributed system is a system that offers

services according to standard rules that describe
syntax and semantics of those services.
 E.g., in computer networks: format, content, and meaning of
messages
 In DS: services specified through interfaces
 Interface Definition Language (IDL): syntax
 Web Service Definition Language (WSDL): syntax
 Semantics? the hard part to specify
 Extensibility: an open DS can be extended and improved
incrementally
 add or replace components

INF5040, ifi/UiO 13
 Scalability

 Scalability in three dimensions:

 Scalable in size: Users and resources can easily be added
 Geographically scalable: Users and resources may lie far apart
 Administratively scalable: The system spans many administrative
organizations

 Scaling up along these dimensions often come with a

performance cost
 Scalability denotes the ability of a system to handle
future scaling up
 Examples:
 Internet: Size, geography and number of administrative organizations has
grown enormously
 Google: scaled over the years to handle O(100) billion queries a month,
expected query time 0.2 secs.

INF5040, ifi/UiO 14
 Scalability Problems

 Problems with size scalability:

 Often caused by centralized solutions

 Problems with geographical scalability:

 traditional synchronous communication in LAN
 unreliable communications in WAN
 Problems with administrative scalability:
 Conflicting policies, complex management, security problems

INF5040, ifi/UiO 15
 Scaling Techniques

Distribution
 splitting a resource (such as data) into smaller parts, and
spreading the parts across the system (cf DNS)
Replication
 replicate resources (services, data) across the system
 increases availability, helps to balance load
 caching (special form of replication)
Hiding communication latencies
 avoid waiting for responses to remote service requests
(use asynchronous communication or design to reduce
the amount of remote requests)

INF5040, ifi/UiO 16
 Fault Tolerance

 Hardware, software and network fail!!

 DS must maintain availability even in cases where
hardware/software/network have low reliability
 Failures in distributed systems are partial
 makes error handling particularly difficult
 Many techniques for handling failures
 Detecting failures (checksum)
 Masking failures (retransmission in protocols)
 Tolerating failures (as in web-browsers)
 Recovery from failures (roll back)
 Redundancy (replicate servers in failure-independent ways)

INF5040, ifi/UiO 17
 Outline

▪ Definition of a distributed system

▪ Goals of a distributed system
▪ Implications of distributed systems
▪ Pitfalls in developing distributed systems
▪ Types of distributed systems

INF5040, ifi/UiO 19
 Implications of Distributed Systems

 Concurrency
 components execute in concurrent processes that read and update
shared resources. Requires coordination
 No global clock
 makes coordination difficult (ordering of events)
 Independent failure of components
 “partial failure” & incomplete information
 Unreliable communication
 Loss of connection and messages. Message bit errors
 Unsecure communication
 Possibility of unauthorised recording and modification of messages
 Expensive communication
 Communication between computers usually has less bandwidth,
longer latency, and costs more, than between independent processes
on the same computer

INF5040, ifi/UiO 20
 Pitfalls When Developing DS

False assumptions made by first time developer:

 The network is reliable.
 The network is secure.
 The network is homogeneous.
 The topology does not change.
 Latency is zero.
 Bandwidth is infinite.
 Transport cost is zero.
 There is one administrator.

INF5040, ifi/UiO 21
 Outline

▪ Definition of a distributed system

▪ Goals of a distributed system
▪ Implications of distributed systems
▪ Pitfalls in developing distributed systems
▪ Types of distributed systems

INF5040, ifi/UiO 22
 Types of Distributed Systems

Type1: Distributed Computing Systems

 Used for high performance computing tasks
 Cluster and Cloud computing systems
 Grid computing
Type2: Distributed Information Systems
 Systems mainly for management and integration of
business functions
 Transaction processing systems
 Enterprise application integration

Type3: Distributed Pervasive (or Ubiquitous)

Systems
 Mobile and embedded systems
 Home systems
 Sensor networks
INF5040, ifi/UiO 23
 Type1: Cluster Computing Systems

 Collection of similar PCs, closely connected, all run same

OS, e.g.:

 A collection of computing nodes + master node

 Master runs middleware: parallel execution and
management

INF5040, ifi/UiO 24
 Type1: Grid Computing Systems

Federation of autonomous and heterogeneous

computer systems (HW,OS,...), several adm
domains

A layered architecture for grid computing systems

INF5040, ifi/UiO 25
 Type1: Cloud Computing

 View: distributed resources as a commodity or utility

 Resources are provided by service suppliers and effectively
rented rather than owned by the end user.
 The term cloud computing capture the vision of
computing as a utility

SalesForce CRM
SaaS
Clients LotusLive

Google App
PaaS Engine
Internet

IaaS
26

INF5040, ifi/UiO 26
 Type2: Enterprise Application Integration

Allowing existing applications to directly exchange

information using communication middleware

Middleware as a communication facilitator in

enterprise application integration

INF5040, ifi/UiO 27
 Type2: Example Communication Middleware: CORBA

Clients may invoke methods of remote objects

without worrying about:
 object location, programming language, operating system
platform, communication protocols or hardware.

X Y Z
invoke Z’s Different
method foo() foo() programming languages
(or object models)
IDL IDL IDL
Common object model

Object Request Broker (ORB) RMI over IIOP

INF5040, ifi/UiO 28
 Type3: Distributed Pervasive Systems

Pervasive systems:
exploiting the increasing integration of services and
(small/tiny) computing devices in our everyday
physical world

(Mobile) Devices: discover the environment (its

services) and establish themselves in this
environment as best as possible.
 Requirements for pervasive applications
 Embrace contextual changes
 Encourage ad hoc and dynamic composition
 Recognize sharing as the default
INF5040, ifi/UiO 29
 Type3: Example: Smart Home System

INF5040, ifi/UiO 30
 Summary

 Distributed systems:
 components located in a network that communicates and coordinates
their actions exclusively by sending messages.
 Goals like resource sharing, distribution transparency,
openness, scalability, fault tolerance and heterogeneity can
be satisfied by distributed systems
 Consequences of distributed systems
 Independent failure of components
 Unsecure communication
 No global clock
 Many pitfalls when developing distributed systems
 Novel applications in pervasive systems: e.g., smart homes

INF5040, ifi/UiO 31