Distributed Operating Systems (Unit 1)

Transparency in DOS

• Access - Hide data representation & resource access.
• Location - Hide the location of the resource
• Migration - Hide the movement of resource and where it is stored.
• Relocation - Resources can move while under use.
• Replication - Create multiple instances of the resources.
• Concurrency - Resource can be shared with other users to be used simultaneously.
• Failure - Hide the failure and recovery of resources from the user,

Pitfalls/Problems in DOS

• Reliable Network - Ensure a lesser number of network failures.
• Secure Network - Communication between 2 systems can not be abruptly interrupted.
• Homogenous Network - Communication must happen regardless of the hardware.
• No change in topology - Constant topology algorithm is unreliable.
• Zero latency - There should be an as little delay between system communication as possible,
• Infinite bandwidth - Sufficient bandwidth for communication.
• Transportation cost - Ingress & Egress is free.
• Single administrator - Have enough administrators for the smooth running of the roles.

Types of DOS

• Distributed Computing Systems
• High-Performance Computing - HPC
• Cluster Computing
  
  
• Grid Computing
  
  
  
• Distributed Information System
• Transaction processing Systems - TPS
  

  
• ACID
• Enterprise Application Integration - EAI
  
  
  
• Nested Transactions
  
  
  
• Distributed Pervasive Systems
• Ubiquitous Systems
• Sensor Systems
  
  
  

Computer Design Concepts

Multiprocessors

• Tightly VS Loosely Coupled
• Bus Based
  
  
  
• NUMA & UMA
• Switch based
  
• Crossbar Switch
  
• Omega Networking
  
  
  

Multicomputer

• Bus-Based
  
  
  
• Switch-Based
  
  
  
• NOS
  
  
  

Design Issues in DOS

• Transparency
• Flexibility
• Reliability
• Performance
• Scalability