- 2075 Chaitra · 4+4 marks
DS vs centralized; layers of transparency
Short answer
Distributed systems are preferred for better price/performance, reliability, scalability, resource sharing, and geographic distribution without a single bottleneck. Transparency layers are network (access and location), system (migration, replication, concurrency, failure), and operational (performance and scaling).
Why is a distributed system preferred over a centralized system? Explain the layers of transparency.
Why a distributed system is preferred
- Better price/performance: many low-cost machines are cheaper than one mainframe.
- Higher performance: work is shared by parallel processing and load balancing.
- Reliability and availability: failure of one node does not stop the whole system; replicas continue the service.
- Scalability: capacity is increased by adding nodes, while a centralised system is limited by one machine.
- Resource sharing: data and devices are shared across sites.
- Geographic distribution: an organisation with branches (a bank) keeps data near its users, with low access delay.
- No single bottleneck or single point of failure.
Layers (levels) of transparency
Transparency can be viewed at three layers.
- Network (access) layer: access transparency (same operations for local and remote resources) and location transparency (names do not reveal location).
- System (resource management) layer: migration/relocation (resources move), replication (copies), concurrency (shared use) and failure (recovery).
- Operational layer: performance (reconfiguration to keep performance) and scaling (the system grows without changes to its structure).
| Transparency | What is hidden | Example |
|---|---|---|
| Access | Differences in data representation and how a resource is accessed | Same call to read a local or remote file |
| Location | Where the resource is physically located | URL does not show the server's place |
| Migration | A resource may move to another location | A process moved to another host without the user noticing |
| Relocation | Movement of a resource while it is in use | Mobile user keeps a call while moving between cells |
| Replication | Existence of several copies | Many mirror servers behind one name |
| Concurrency | Sharing of a resource by several users | Two users editing a database safely |
| Failure | Failure and recovery of a component | Request is served by a backup when a server crashes |
| Persistence | Whether the resource is in memory or on disk | Object is loaded from disk when needed |
| Performance / scaling | System reconfigures to keep performance or grow | Load balancer adds servers |
✓ Answer checked by Claude Opus 5.5 (high effort)
Related questions
- Centralized vs distributed; design issues · Distributed System, chapter 1
- Transparency properties of DS · Distributed System, chapter 1
- Why NOS preferred over DOS · Distributed System, chapter 3 · asked 2×
- Centralized vs Ricart-Agrawala algorithm · Distributed System, chapter 6
- RPC vs RMI; modern RPC transparency · Distributed System, chapter 2
- Autonomous vs distributed system · Distributed System, chapter 1
- Why serial transfer is preferred over parallel · Instrumentation II, chapter 3
- T/F: ACSR preferred over AAAC · Transmission and Distribution System Design, chapter 4
- File system layers; linked list vs index · Operating System, chapter 5
- Distributed system in big data example · Big Data Technologies, chapter 1