BCIT 1105 - Introduction to

Operating Systems

Distributed Operating Systems

Distributed Systems
1. A distributed system is a collection of processors
without shared memory or clock, each with its own
local memory.
2. Nodes communicate through networks like high-
speed buses or the Internet.
3. Components in a distributed system are
autonomous computers connected by a
communication medium.
4. Applications execute across multiple systems, with
processes communicating to achieve tasks.
5. To users, a distributed system appears as a single
computer, though it is composed of multiple
machines.
6. Each node views others as remote, managing its
resources locally while sharing resources across the
system.
Distributed O/s
Not all operating systems are on a single CPU
The nature of the distribution varies widely
Types of Distributed Computes
i. Multiprocessors
Multiple CPUs on a single bus
Current trend in chip and system design
ii. Multi-computers
Many independent computers connected by a
i. switching fabric
ii. Memory is not shared
iii. No bus contention
iv. Contention for switching fabric
v. Distributed systems (and the Global Grid)
Benefits of Distributed Systems
1. Resource Sharing
i. Mechanisms for sharing files at remote sites,
processing information in distributed databases,
printing remotely, and using specialized devices.
ii. Enables access to remote supercomputers and
other hardware for optimized performance.

2. Computation Speedup
iii. Allows partitioning of tasks into sub-computations
that run concurrently across sites for faster
processing.
iv. Enables load sharing or job migration, moving tasks
from overloaded sites to those with less demand.
Benefits of Distributed Systems
3. Reliability
i. If one site fails, others can continue, ensuring the
system remains operational.
ii. Failure of one large installation doesn’t impact the
functioning of other parts of the system.

4. Communication
iii. Connects multiple sites for collaborative work and
information exchange.
iv. Facilitates long-distance project collaboration,
enabling activities like file sharing, remote logins,
and email exchange.
Types of Network-based
Operating
. Systems
Two main types of network-oriented OS: Network
Operating Systems (NOS) and Distributed
Operating Systems (DOS).
Network Operating Systems (NOS)
i. Simpler to implement, but users must log into
remote machines or transfer data manually.
ii. Examples: General-purpose OS like Windows,
Linux; also includes mobile OS like Android and
iOS.

Distributed Operating Systems (DOS)

iii. Users access remote resources seamlessly, as if
local.
iv. Manages data and process migration
automatically across sites.
Components of Distributed O/S
1) Data Migration:
Transfer the entire file or only necessary portions to the
local site (e.g., Site A to Site B).

2) Computation Migration:
Transfer computation to the site where data resides,
process it locally, and return the results.
- Example: Summarizing large files at their respective
sites.

3) Process Migration:
A logical extension of computation migration is process
migration.
Execute a process at a site different from where it was
initiated.
- Parts of a process may execute across multiple sites.
Features Distributed Systems
1. Multiple autonomous processing
- Composed of several independent components,
each with processing ability.
- No master-slave relationship, excluding centralized
mainframe.
2. Information exchange over a network
- Network connects autonomous network processing.
3. Processes interact via intercommunication
- Systems classified into shared memory and non-
shared memory.
- High cooperation involves separate computers with
distributed shared memory.
 Distributed Systems
Basic Issues in Network Design

1) Naming and Name Resolution: How do two

processes locate each other to communicate?

2) Routing Strategies: How are messages sent through

the network?

3) Packet Strategies: Are packets sent individually or

as a sequence?

4) Connection Strategies: How do two processes send

a sequence of messages?
 Distributed Systems
Basic Issues in Network Design

- Hostname resolved into host-id for efficiency

- Names are consistent and globally meaningful
- Global access ensures programs run and access
services seamlessly
Single or multiple physical paths from source to
destination
- Messages routed through chosen paths based on
network topology
- Critical for efficient and reliable communication

System components can be administered universally

- Management tools perform consistently across
configurations
- Ensures uniformity and simplicity in network
operations
 Distributed Systems Models
Mini Computer Model
It may be used for resource sharing such as sharing
information in databases or different machines.
Distributed Systems Models
Workstation Model
 Distributed Systems Models
Workstation server model
Advantage of the model as compared to work station model
▪ It is cheaper
▪ Easier to maintain
▪ Use of file server makes it possible for users to use any work
station but still access these files.
 Distributed Systems Models
Processor pool model
Each processor in the pool has its own memory.
▪ It allows for better utilization of the availability
processing power of the system.
▪ It is not suitable for high performance interactions
application especially graphics because of slow
communication.
 Advantages of in Distributed
1. Fault Tolerance
Systems
- System continues to function even if a component
fails.
2. Flexibility
- Supports diverse and specialized components; easy
to add, move, or remove parts.
3. Easy to Extend
- Increase processing, storage, or power by adding
components.
4. Easy to Upgrade
- No need to replace the entire system, minimizing
costs and disruption.
5. Local Autonomy
- Individual domains manage their own processing,
storage, and resources.
 Disadvantages of Distributed
1. Difficult to manage security:
Systems
- Centralized systems are easier to secure as control is from a
single point.
- Distributed systems require complex security, administration,
and user support due to dispersion.
2. Lack of skilled support:
- Equipment and software in distributed systems often come
from different vendors.
- Hard to find personnel with diverse skills for effective system
management.
3. Significant complexity:
- Distributed systems are inherently more complex than
centralized systems.
4. Synchronization issues:
- Problems arise in ensuring processes are synchronized across
the system.
5. Data consistency challenges:
- Maintaining consistent data across distributed nodes can be
difficult.
 Types Distributed Systems
1. Tightly Coupled System
- Single system-wide primary memory shared by all processors.
- Processors communicate through shared memory.
- Referred to as parallel systems.

2. Loosely Coupled System

- Processors are geographically distributed.
- Communication through message passing over a network.
- Each processor has its own local memory.
Tightly vs. Loosely Coupled Systems

Tightly Coupled System:

- Processors share a single primary memory.
- Limited by bandwidth of shared memory.

Loosely Coupled System:

- Processors can be geographically distributed.
- Each processor has its own local memory
 Types Distributed Systems
Tightly Coupled System
 Types Distributed Systems
Loosely Coupled System
 Issues in designing a
distributed O/S
1. Resources are physically separated.

2. No common clock among processors.

3. Delivery of messages may be delayed or lost.

4. Transparency: Ensuring a single system image is given to

users
 Transparency in Distributed
1. Designed to conceal from users that they operate over a
Systems
wide area.
2. Provides an illusion of a single desktop environment.
3. Aspects of transparency include:
- Global names.
- Uniform services across the system.
i. Location Transparency; Hides the physical or network
location of resources, allowing access without knowing
actual locations.
ii. Access Transparency; Hides differences in data
representation or communication methods, enabling
uniform access to resources.
iii. Replication Transparency; Hides the existence of
resource replication to provide fault tolerance and
performance, ensuring consistency.
iv. Name Transparency; Hides the details of resource
identifiers, allowing users to use familiar names for
resources
 Transparency in Distributed
v. Systems
Migration Transparency: Users unaware of resource/job
relocation

vi. Concurrency Transparency: Efficient handling of

resource sharing.

vii. Failure Transparency: System resilient to individual

failures.
 Distributed File Systems
Goal: provide common view of centralized file system, but
distributed implementation.
i. Ability to open & update any file on any machine on
network
ii. All of synchronization issues and capabilities of shared
local files
Naming – mapping between logical and physical objects.
A transparent DFS hides the location where in the network the
file is stored.
Location transparency – file name does not reveal the file’s
physical storage location.
File name denotes a specific, hidden, set of physical disk
blocks.
Convenient way to share data.
Location independence – file name does not need to be
changed when the file’s physical storage location changes.
Better file abstraction.
Promotes sharing the storage space itself.
Separates the naming hierarchy from the storage-devices
hierarchy.
 Distributed File Systems
A special case of distributed system
Allows multi-computer systems to share files
Sharing devices
Special case of sharing files
E.g.,
NFS (Sun’s Network File System)
Windows NT, 2000, XP
Andrew File System (AFS) & others …
 Distributed File Systems
Summary of Distributed File Systems
There are a number of issues to deal with:
i. what is the basic abstraction?
ii. naming
iii. caching
iv. sharing and coherency
v. replication
vi. performance
vii. workload
No right answer! Different systems make different
tradeoffs…