JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD B.Tech III ADVANCED OPERATING SYSTEMS Previous Question Paper (February - 2022 - R18)

The design of a distributed operating system (DOS) involves addressing several fundamental challenges to ensure transparency, efficiency, and reliability across interconnected nodes:

• Transparency: The system should hide the distributed nature of the hardware and software from the user, providing location, migration, replication, and concurrency transparency.
• Reliability and Fault Tolerance: Since nodes can fail independently, the system must detect failures and recover gracefully through redundancy, checkpointing, or atomic transactions.
• Scalability: The architecture must handle growth in the number of nodes or users without significant performance degradation, often requiring decentralized control and avoiding bottlenecks.
• Communication Models: Choosing between message passing or Remote Procedure Calls (RPC) is critical for inter-process communication, balancing latency and ease of programming.
• Resource Management: Efficient scheduling and load balancing are required to distribute tasks across nodes to maximize resource utilization.
• Synchronization and Deadlock: Managing distributed mutual exclusion and preventing deadlocks in a system without global state requires complex algorithms like token passing or logical clocks.

Vector clocks and logical clocks (Lamport clocks) are mechanisms used to order events in distributed systems where a global physical clock is unavailable.

While logical clocks only track the happens-before relationship in one direction, vector clocks provide a complete representation of causal history, allowing the system to detect if two events are concurrent by comparing the vector entries.