Distributed Systems

A distributed system is defined as a collection of independent computers that appears to its users as a single, coherent computer system. The key characteristic is that the multiple machines work together transparently, so users perceive them as one unified system rather than separate entities.

Data sharing is a primary advantage of distributed systems over standalone PCs. In a distributed system, multiple users can access and share the same data files and resources across different machines, which enables collaboration and reduces data redundancy. This is impossible with isolated PCs.

Distributed systems are more flexible than centralized systems because they can be expanded incrementally, allow resource sharing across different locations, and can adapt to failures by redistributing work. The distributed architecture enables multiple configurations and easy scaling compared to rigid centralized setups.

Fault tolerance is the ability of a system to continue operating properly even when some components fail. In distributed systems, if one machine fails, the system can continue running tasks by redistributing the work to remaining operational machines, ensuring continuous service without interruption.

Location transparency is a key property of distributed systems where users cannot distinguish whether a resource (like a file, printer, or service) is located on their local machine or on a remote computer. The system handles all the details of accessing remote resources transparently, making location irrelevant to the user.

Distributed systems are specifically designed to enable multiple users to work together by sharing resources, data, and processing power across different machines. Unlike parallel systems (which focus on simultaneous processing) or linear systems, distributed systems emphasize collaboration and coordination among users across a network.

Scaling refers to the ability of a system to expand its capacity (by adding more resources, computers, or processing power) without disrupting ongoing user activities. Good scaling allows growth in size and capability while maintaining existing operations and services.

RPC (Remote Procedure Call) is a protocol that allows a program to execute a procedure or function on another computer (remote server) as if it were a local call. The programmer doesn't need to explicitly code the details of network communication - the RPC mechanism handles the transparent communication between local and remote systems.

Microkernels are designed to be minimal, moving most OS services to user space, which makes them highly modular and flexible compared to monolithic kernels.

X.25 is a packet-switching network protocol that is connection-oriented, requiring a virtual circuit to be established before data transmission.

Leslie Lamport defined the 'happened-before' relation in his seminal 1978 paper, which is fundamental to logical clocks and ordering events in distributed systems.

Synchronization is significantly more complicated in distributed systems due to the lack of a global clock and the presence of network latency.

Physical clocks are hardware clocks that must not only be the same but must also not deviate from real time by more than a certain amount. They have bounded drift rates. Real clocks and logical clocks are different concepts - logical clocks track event ordering without reference to real time. The question tests understanding of physical clock precision requirements.

In Lamport's logical clock theory, the notation a -> b signifies that event 'a' happened before event 'b' in the causal ordering of the system.