Distributed Systems: Definition, Pros & Cons

Definition of a Distributed System

• A distributed system involves multiple connected CPUs working together.
• It appears to users as a single, coherent system, although it consists of independent computers.
• Parallel machines and networked machines are examples of distributed systems.

Advantages and Disadvantages

• Advantages include the possibility of communication, resource sharing, a good price-performance ratio, reliability, scalability, and potential for incremental growth.
• Disadvantages include the need for distribution-aware programming languages, operating systems, and applications, as well as essential network connectivity, security, and privacy considerations.

Transparency in a Distributed System

• Access transparency hides differences in data representation and how a resource is accessed.
• Location transparency hides where a resource is located.
• Migration transparency hides that a resource may move to another location.
• Relocation transparency hides that a resource may be moved to another location while in use.
• Replication transparency hides that a resource may be replicated.
• Concurrency transparency hides that a resource may be shared by several competitive users.
• Failure transparency hides the failure and recovery of a resource.
• Persistence transparency hides whether a software resource is in memory or on disk.

Open Distributed Systems

• Open distributed systems offer services described a priori
• Syntax and semantics are known through protocols.
• Services are specified via interfaces, offering benefits like interoperability and portability.
• Open systems should evolve over time, be extensible to new functionality, and separate policy from mechanism.

Scalability Problems

• Centralized services, such as a single server for all users, can cause scalability problems.
• Centralized data, like a single on-line telephone book, can limit scalability.
• Centralized algorithms, such as doing routing based on complete information, can hinder scalability.

Scaling Techniques

• Principles for good decentralized algorithms include no machine having complete state.
• Decisions should be based on local information, and a single failure should not bring down the entire system.
• No global clock is needed.
• Techniques include asynchronous communication, distribution, and caching and replication.

Module 3: Distributed Systems History and OS Models

• Minicomputer model with early networks: each user has a local machine and can fetch remote data.
• Workstation model uses Sprite, allowing processing to migrate.
• Client-server Model: Users have a local workstation, and powerful workstations serve as servers.
• Processor pool model: Terminals are Xterms, and a pool of backend processors handles processing.

Distributed System Models (contd)

• Cluster computing systems/Data centers consist of LAN with a cluster of servers plus storage (e.g., Linux, Mosix).
  • Used by distributed web servers, scientific applications, and enterprise applications.
• Grid computing systems: a cluster of machines connected over a WAN, such as SETI @ home.
• WAN-based clusters/distributed data centers include Google and Amazon.
• Virtualization and data centers are used.
• Cloud Computing is also implemented.

Emerging Models

• Distributed Pervasive Systems are emerging with smaller nodes that have networking capabilities.
  • Computing is everywhere,
  • Home networks include TiVO and Windows Media Center and mobile computing includes smart phones, iPads, and car-based PCs.
• Sensor networks and health-care/personal area networks are utilized.
• Sustainability is considered as a design goal.

Uniprocessor Operating Systems

• An OS acts as a resource manager or an arbitrator and manages CPU, I/O devices, and memory.
• The OS provides a virtual interface that is easier to use than hardware.
• Structure of uniprocessor operating systems can be monolithic (e.g., MS-DOS, early UNIX) with one large kernel.
• Layered design decomposes functionality into N layers, where each layer uses services of layer N-1 and implements new service(s) for layer N+1.

Microkernel Operating Systems

• Microkernel architecture features a small kernel, and user-level servers implement additional functionality.
• There is no direct data exchange between modules.

Distributed Operating System

• It manages resources in a distributed system seamlessly and transparently to the user.
• The system appears like a centralized OS but operates on multiple independent CPUs.
• It provides transparency in location, migration, concurrency, and replication.
• It presents users with a virtual uniprocessor.

Types of Distributed OSs

• DOS (Distributed Operating System) is a tightly coupled operating system for multi-processors and homogeneous multicomputers, with the goal of hiding and managing hardware resources.
• NOS (Network Operating System) is a loosely coupled operating system for heterogeneous multicomputers (LAN and WAN), offering local services to remote clients.
• Middleware is an additional layer atop of NOS, implementing general-purpose services and providing distribution transparency.

Multiprocessor Operating Systems

• It is like a uniprocessor operating system that manages multiple CPUs transparently to the user.
• Each processor has its own hardware cache, and the system maintains consistency of cached data.

Network Operating System

• Employs a client-server model.
• Minimal OS kernel.
• Additional functionality as user processes.

Middleware-based Systems

• General structure of a distributed system as middleware.