Fundamentals of data communication in distributed systems: Protocols, architectures and challenges
Objectives
After completing this reading, you should be able to;
?
Computing and telecommunication evidently depend heavily on data transmission. This is the underlying principles of how communication systems are possible. Data communication can simply be said to be the transfer and flow of data from one system to another. This can be two or more geographically dispersed systems otherwise known as workstations. Communication can be seen in the interaction of a weather app(client) and weather API(Server) to provide weather information of a given city.
?
What is a distributed system?
Distributed system is a unique way of implementing application environment to ensure availability, scalability, reliability and extensibility. Decentralized systems are often confused with distributed system but differ in that no component is solely responsible for the decision making. A distributed system is a computing environment where several processes run simultaneously on different machines and communicate through coordinated actions on ?a network to appear to end users as a single system. An example of a distributed system is Netflix that delivers reliable, scalable and high-performance services to a global audience.
?
Communication protocols in distributed systems.
In distributed systems, communication protocols define how computers identify one another on a network. Communication protocols are otherwise known as standards that ensure that all hardware and software can communicate as long as they conform to the same specification. They include
Remote Procedure call(RPC): This is a software communication protocol in distributed systems that allows one program to request a service from another program located remotely similar to how a local system would. RPC allows the execution of a procedure by one program in the address space of another while abstracting away the complexity of network communication
Transmission Control Protocol(TCP) / Internet Protocol(IP) : TCP protocol ensures reliable data transfer by way of dividing messages into small chunks called IP packets from the source computer and reassembling them in the destination computer. These packets are made up of both headers which contain information about the destination computer and a payload containing the actual data being transferred. Internet protocol otherwise known as IP is responsible for handling the address of the destination computer, hence ensuring that packets are delivered to the destination computer. These form the foundational protocol suite for internet communication responsible for data communication and routing.
Message passing protocol: Message passing in distributed system is a medium of communication between nodes to allow the transfer of information and coordinate their actions. It enables the communication of different components of a system thereby achieving the goal of synchronization and data sharing. An example of message passing protocol can been seen in Software as a service systems such as Gmail and outlook that ensures reliable delivery of messages from your device to a recipient mailbox
Data Serialization Protocols: Data serialization are methods and standards of converting ?complex data structures into formats that can be easily stored, transmitted and reconstructed. These protocols are integral to communication in distributed systems, APIs and interservice communication. Examples of Data Serialization protocol are JavaScript object Notation(JSON), Extensible Markup language that ensures readability and compatibility.
WebSocket protocol: WebSocket protocol is a communication protocol based on client-server communication channel. It is a full-duplex communication protocol otherwise known as a bi-directional communication that allows data to be sent to and from a server. It is especially useful for applications that require low latency and continuous communication. An example use case of WebSocket protocol is seen in real time chat applications like Slack, Discord and WhatsApp.
Peer-to-peer(P2P) Communication Protocol: P2P process in a network deals with a communication structure where each node acts as a client and server. It enable the communication between devices in a decentralized manner thereby bypassing centralized servers. It is used in various applications such as file sharing like BitTorrent, distributed systems, real-time communication and blockchain communication.
User Datagram Protocol(UDP): UDP is a connectionless and unreliable data communication protocol used in time sensitive transmissions such as video playback and DNS lookup. Its eliminate the process of forming a connection, thereby resulting in a speedier data transfer. UDP can cause packets to get lost and therefore does not guarantee delivery. These protocols are mostly used in video game developments particularly multiplayer games because it prioritizes speed and low latency over reliability.
Hyper Text Transfer Protocol(HTTP) and Hyper Text Transfer Protocol Secure(HTTPS): This is a request-response protocol in client-server computing model forming the foundation of data exchange on the web. It supports the transfer of hypertext documents on the world wide web. HTTP and HTTPS operates over TCP and supports various request methods such as POST, GET, PUT, PATCH and DELETE to interact with resources. The major distinction between HTTP and HTTPS lies in the security in communication between the client and server.
Each of these protocols has its strength and weaknesses and is chosen based on the specific requirement of the system. The features of a communication protocol can influence architectural model. For example bidirectional communication can lead to adoption of real-time or even-driven architecture.
?
领英推荐
Software Architecture of distributed systems
Architecture in software systems can be described as an overview of the system and how components are organized to ensure efficiency in communication. The emergence of cloud and serverless computing has influenced the way software systems are designed as we see a shift from monolithic architecture to a more distributed architecture. In monolithic architecture, ?components of a system are tightly coupled and run on a single server, thereby resulting in a single point of failure. Distributed architecture offers a more flexible and scalable approach. As there is an increasing need for more reliable and performant systems, understanding the principle and benefit of distributed systems is crucial for architects and developers to build systems that meet the need of customers. Below are some of the architectural models of distributed systems.
Layered Architecture: In layered architecture, components are arranged into hierarchical layers based on specific functions and responsibilities thereby ensuring reliability. It separates components into layers as a way to promote separation of concern and manage complexity. In layered architecture, request goes from top down and response goes from bottom up. This helps to keeps things orderly and ensure independent modification of layers.
Object based architecture: Object based architecture emphasizes encapsulating data and behavior within objects. This approach is widely used in object oriented programming and foundational to many modern software design principles. It centers around loosely coupled objects where each object can interact through an interface otherwise known as a connector. At its core, communication between objects happen through method invocation(RPC). Components in object based architecture are less structured with each component acting as an object and connectors as RPC or RMI.
Data centered architecture: Data centered architecture focuses on the central management and utilization of data. This is considered a data repository either active or passive. The system is designed around data management, storage and retrieval. This central repository can be just a simple database like SQL. An example of data centered architecture is the Data Warehouse Architecture use in business Intelligence that focuses of data storage, retrieval and analysis.
Event-based Architecture: In event based architecture, the entire system communicates through events. Events are occurrences or changes in states that trigger a response, the system gets a notification and components are able to subscribe to these events to access information. Sometimes, these events are data but can also be URL to resources. This approach decouples the components of a system thereby allowing asynchronous communication and enabling flexibility, scalability and responsiveness.
Client-server Architecture: Client-server architecture is a foundational model in distributed systems where the system is divided into two distinct components: client and server. The server is where data is processed while the client offers the interface where the user interacts with the service and other resources. This accounts for the single point of failure but ensures a more stable and secure system at the cost of speed.
Peer-to-peer(P2P) Architecture: In peer-to-architecture, each node can perform the job of both a server and a client. They collectively contribute resources and services to the network. Each node is either a server or a client at any given time once it joins the network. Whichever node request a resource is called a client and whichever provides a service is called a server. An example of a peer-to-peer architecture is the bitcoin network that emphasizes security, distributed transactions without relying on a central authority.
Service-oriented Architecture(SOA): SOA is a method in software development that utilizes several components called service each providing a different service to make up a complete business application. Each components provides a distinct business capability and can communicate across platform and languages. Developers either reuse components in different services or use a combination of services to perform complex tasks. An example of Service-Oriented Architecture is the amazon web service, where each service interacts through a well-defined interface(API) to deliver complex solutions in a modular and scalable way.
Microservice Architecture: ?Microservice is an architectural style that structures an application as a collection of two or more services that are independently deployable and loosely coupled. These service are organized around business capabilities and owned by a single small team. Netflix is a good example of a microservice architecture with different microservices working together to manage the various aspects of video streaming.
For businesses to thrive in today’s volatile, complex and ambiguous world, IT must aim to deliver software systems that are efficient, reliable and easily scalable. This might result in hybrid system that combine various architecture to achieve customized business logic.
?
Challenges of Distributed systems
While distributed systems offer advantages such as such scalability, fault tolerance and flexibility, these come at a cost to performance and several other challenges not limited to complexity. These challenges arise from the need to manage resources, handle communication, and ensure consistency across multiple nodes in the system. Here are some of the challenges.
Communication complexity: In distributed systems, coordinating actions across nodes requires effective communication, However latency, packet loss and communication volume can account for the level of complexity in the application design, performance and responsiveness.
Concurrent resource access: Concurrency is the ability to process data parallelly on different nodes. Communication and synchronization among nodes can pose a challenge. The system needs ?have sufficient fault tolerance mechanism mitigate system failure.
Security concern: Communication across networks, data transmission and several access controls further expands the attack surface of the system providing critical points of vulnerability.
Scalability: As the system scales to accommodate for increased load and manage communication across disperse locations, the design must aim for performance without degradation. This can pose a challenge because several level of abstraction needs to accounted for to ensure ease of scalability.
Development and bug reproduction: With geographically dispersed nodes, communication happens across different nodes and services. This a creates a ?communication path for requests that can be challenging to manage and debug failures in production.
?
IT has seen emergence in trends such as edge computing that focus on decentralized computation and storage while also ensuring proximity to data source. AI driven systems have also entered the scene with trends of incorporating machine learning into system decision making and data processing, hence enhancing data transformation and system capability. There are also zero trust systems that focus on security by assuming no implicit trust for users and devices within and outside a network perimeter. These have contributed to the use of specialized protocols and customized architectures that focus on more decentralization and specialization of services across various networks.