Distributed systems

Distributed systems

       Nowadays we no longer consider computers just as standalone machines that help us work faster and increase our productivity. Instead, computers are getting more and more interconnected, sharing their resources and computing power.

    Just like a single computer consists of many dedicated hardware devices, each with a separate contribution to the whole system (for example – a processor, fast speed operative memory, storage memory, graphics card, data input devices and so on), a distributed system employs a similar architecture, but on a more global scale. They are a collection of autonomous computers, connected through network architecture, coordinating their activities and sharing their resources in such a manner that end users perceive the system as a single facility. Distributed systems could be allocated within a whole building, why not city, country or even continent. The introduction of distributed systems in an enterprise increases scalability, performance, fail-safe operability and reparability, as data is not allocated entirely on one unit, but shared among many. Thus, failure of a node does not critically affect overall performance.

      There are, however, some key differences between single and distributed systems. Apart from, of course, the physical separation of the different computers, there is no global clock. The explanation for this is pretty simple, actually: each computer of the system has their own clock that runs at its unique pace, pretty similar to the others, but still different. Therefore, there cannot be a unique global time for the whole system and only local clock is taken into account.

         The other distinctive feature of distributed systems is network delay, or network latency. Due to the fact that the physical distance between nodes could be miles and more, and network links could be switched and routed through many servers, latency is inevitable and needs to be respectfully taken into account.

      Depending on a distributed system’s design and settings, different computers could work concurrently (or simultaneously) towards the completion of a single task. This is not a set rule and can vary from system to system.