A distributed system is a system in which a large number of computers are connected via a network and are operating while sharing work.
In recent years, distributed systems have become the mainstream of system configurations because the load on each machine is small and the system does not go down even if it breaks down.
This time, we will introduce the outline of the distributed system, its advantages and disadvantages, and the “autonomous distributed system” that has evolved from the conventional distributed system.
table of contents
- 1. What is a distributed system?
- 2. Form of distributed system
- 3. Advantages and disadvantages of distributed system
- 4. “Autonomous decentralized system” that thinks and acts on its own
- 5. Expectations and possibilities for an ever-changing self-sustaining distributed system
What is a distributed system?
Distributed systems are systems in which work is shared among multiple computers connected by a network. Processing is distributed by connecting multiple terminals in a network environment.
● Characteristics of distributed system
The computers used in a distributed system are independent, but from the user’s (external) perspective, the whole system looks like a coherent system. In reality, many terminals cooperate to share the processing, but it is not visible from the outside that the process is shared and processed.
Until now, it was mainly used by companies and schools, but in recent years, online-connected systems such as the Internet and intranets have been mainly used in various places. For example, the client-server system (CSS) is a typical model for distributed systems. The server is in charge of specific processing such as mail server and data server, and the client charges the server for necessary processing.
● Antonym of distributed system “centralized system”
The system that contrasts with the distributed system is the “centralized system”. In a centralized system, all processing is centralized on the mainframe, and on the client-side, only user I / O processing is performed. The thin client is one of the representative models of centralized systems.
The centralized system is capable of processing large volumes of data, is easy to manage operations and security, and is highly reliable. It has been used before distributed systems and was the mainstream system until the 1980s. Centralized systems are still used in some systems where reliability is important.
However, the centralized system has the disadvantage that the load on the mainframe is heavy, and if the mainframe should fail, the system will stop and it will take a lot of time to recover. In addition, due to the high introduction cost, distributed systems have become the mainstream in recent years.
Form of distributed system
Distributed systems can be broadly divided into vertical distributed systems and horizontal distributed systems. Here, we will explain the difference between the two forms.
● Vertical distribution
It is a distributed system in which computers with different roles are arranged in hierarchical order. It is configured so that multiple medium-sized and small-sized terminals hang around a large computer such as a mainframe. The central large computer and the medium and smaller terminals have different roles and are similar to a centralized system, but since the sub terminals also perform distributed processing in cooperation, the load on the main large computer can be reduced.
The client-server system used in corporate information systems is a typical vertical distributed system, consisting of medium-sized and small terminals in sales offices and branches arranged around a large computer at the head office. It is a model.
● Horizontal distribution
It is a distributed system that connects multiple computers with almost the same role on an equal footing. There is no hierarchical relationship between computers, they are composed of equal systems, and processing is distributed after sharing resources such as databases and programs. There are two types of horizontal distribution systems: a “horizontal load balancing system” which distributes the processing of the same functional content and averages the load, and a “horizontal function distribution system” that distributes the processing for each function.
A distributed database is a typical model of a horizontal distributed system, in which multiple data servers are connected via a network and treated as if they were a single database.
Advantages and disadvantages of distributed systems
Distributed systems have the advantage of being able to distribute the load, but they also have the disadvantage. Introducing the advantages and disadvantages of distributed systems.
<Since it is distributed, the burden on one computer is small>
Since the processing is shared by multiple computers, the load is small and the processing is completed quickly. In addition, by distributing the load, it is possible to perform large-scale processing that cannot be done by one unit.
<Even if a partial failure occurs, the effect is small>
Since processing is performed by multiple computers, the entire system will not go down even if one fails. The load on other computers will increase, but the system will continue. The range affected by the failure of one device is small, and if the failed terminal is repaired or replaced, the process can be restored.
Since it is easy to increase or decrease the number of terminals, you can expand the system or add the desired functions according to your purpose. For example, you can add terminals to speed up processing or make it more fault-tolerant. If you put a spare terminal on standby in the system in advance, you can immediately replace it with a substitute terminal in the unlikely event that the terminal breaks down.
A centralized system requires a large computer at the top, while a distributed system can only be configured with a collection of small terminals. If it is a small terminal, the price is low, so the cost at the time of introduction can be suppressed. You could start with a low-cost system and gradually buy and expand your device.
<Complicated system management>
In a distributed system, the number of computers that need to be managed increases, which complicates operational management. You should update all your computers with software and monitor your network connection for problems.
<Difficult to take security measures>
In a distributed system, all computers require security measures. Therefore, it takes time and money to take overall security measures.
“Autonomous decentralized system” that thinks and acts on its own
As an evolutionary system of distributed systems, there are “autonomous distributed systems” used in fields where reliability is important. What are the advantages of a self-sustaining distributed system compared to a conventional distributed system?
● What is an autonomous decentralized system?
An autonomous decentralized system is a system that does not have a large computer like a mainframe in the center, and all terminals act autonomously while operating as one system as a whole.
The work is shared by multiple computers, but usually, the processing shared by each is performed independently. If there is a process that requires cooperation, a request is sent to the system and the whole functions in cooperation.
Since each is an independent system, even if a problem occurs somewhere, it will not affect the system and you can replace, add or delete computers as they are. It is evaluated as a highly reliable system because the risk of the entire system going down is low.
● Application example of self-sustaining distributed system
Due to its high reliability, self-sustaining distributed systems are mainly used in fields that must be kept in operation, such as steel production control and factory automation processes. Autonomous decentralized systems are also used in unstoppable infrastructure fields such as JR companies and private railway IC card tickets, which perform a large amount of high-speed real-time processing.
The Internet itself is considered to be a typical model of a self-sustaining distributed system. This is because the Internet operates in cooperation with autonomous and independent networks without any particular administrator, and is adapted to the ever-changing environment.
Expectations and possibilities for an ever-changing self-sustaining distributed system
In recent years, it has become commonplace for computers to be connected to networks.
From now on, IoT (Internet of Things) will spread, and, certainly, systems that are always in operation and do not have to stop are needed everywhere.
Therefore, there will be an increasing demand for self-sustaining distributed systems that are highly reliable, have the high processing power, and can change quickly and flexibly to achieve overall optimization.
■ To find a subcontractor for system development, use ” Order Navi “
Have you already decided on a system development company?
Order Navi is a supplier selection support service that specializes in system development.
From more than 1,500 development companies nationwide, we will search for the best supplier on your behalf.
It is completely free from consultation to introduction (no-contract fee is required).
With the shortest one-day speed introduction, you will not have to wait even for urgent projects.
If you ever want to know about similar things, check out the Facebook page Maga Techs