- 2072 Chaitra · 2+6 marks
Centralized vs distributed; design issues
Short answer
Centralized system has one central computer with single control and single point of failure; distributed system has independent networked nodes with partial failures and better scalability. Design issues include transparency, communication, naming, process and resource management, consistency and replication, synchronization, fault tolerance, scalability, security, and heterogeneity.
Differentiate between centralized and distributed system. Explain the design issues related to distributed system.
Centralized versus distributed system
| Point | Centralized system | Distributed system |
|---|---|---|
| Nodes | One central computer, users use terminals | Many independent computers on a network |
| Control | Single point of control | Control and data are spread over nodes |
| Failure | Single point of failure; whole system stops | Partial failure; remaining nodes continue |
| Scalability | Limited by the central machine | Grows by adding nodes |
| Cost | Costly mainframe | Cheaper commodity machines |
| Resource sharing | Local only | Across the network |
| Response time | Falls when the load grows | Better through load sharing |
| Clock and memory | Shared memory and one clock | No shared memory, no global clock |
| Security and management | Easier | Harder |
| Example | Mainframe with terminals | Web, cloud |
Design issues
- Transparency: what to hide (access, location, replication, failure...) and at what cost.
- Communication: choice of message passing, RPC, RMI, group communication and protocols.
- Naming: unique, location independent names and an efficient name service (DNS).
- Process and resource management: scheduling, load balancing, thread use, migration.
- Consistency and replication: keep copies consistent while providing speed.
- Synchronisation and time: no global clock; need logical clocks, mutual exclusion and election.
- Fault tolerance: detection, masking and recovery.
- Scalability and performance: avoid bottlenecks, use caching.
- Security: authentication, authorisation and secure channels.
- Heterogeneity and openness: standard interfaces, middleware.
✓ Answer checked by Claude Opus 5.5 (high effort)
Related questions
- Autonomous vs distributed system · Distributed System, chapter 1
- DS vs centralized; layers of transparency · Distributed System, chapter 1
- Centralized vs Ricart-Agrawala algorithm · Distributed System, chapter 6
- Network OS vs distributed OS · Distributed System, chapter 3
- Models of distributed system · Distributed System, chapter 1
- True distributed system; middleware organisation · Distributed System, chapter 3
- Short note: Distributed system implementation · Object Oriented Analysis and Design, chapter 4
- Medium network design issues and proxy servers · Internet and Intranet, chapter 5
- Lumped vs distributed analysis, applications, protocols · RF and Microwave Engineering, chapter 1
- Transmission line issues in high speed design · Instrumentation II, chapter 7