In short, network computing is the sharing of resources and distributing of processing across devices in a network. Each segments a variety of models (such as client-server, peer-to-peer and cloud computing, grid computing, distributed computing, edge computing and thin clients – which basically do little other than run almost everything from a central server). The appropriate model depends on where your work needs to be done, be it by the user, in a data center or spread across numerous machines.
Meaning and Core Idea
Network Computing is a process that allows a group of machines to exchange resources like files, applications, printers, etc and Internet access. It also allows for distributing processing tasks to many devices. For instance, in the realm of cloud computing, complex calculations take place on remote servers, thereby lightening up the load getting rid of burden on the local devices.
The concept behind all of this is that of client-server model, a distributed application architecture in which tasks are divided between service providers (servers) and service seekers (clients). Client and server can be run independently on separate machines that are connected to a network, or on the same machine.
The Main Models Compared
The Main Models compared (Original comparison)
This is an editor’s table summarizing the location of work in each of these models, as described in published definitions of the glossary terms.
| Model | Where the processing happens | Best for | Trade-off |
| Client-server | Has one server that serves multiple clients | Business applications and Web services | Client-server can be a single point of failure or neck for server |
| Peer2Peer | devices share resources between devices, without a server to manage and secure the locations, devices, and so on | Centralized sharing | Easier to manage and secure. |
| Cloud | On demand | elastic workloads & storage | Depends on connectivity and continuous costs |
| Matrix | Multiple machines computing multiple tasks (processes) | Heavy computation | Complex to coordinate. |
| Cooperating distributed systems | Independent computers working on various tasks | Large and Multi-part systems | Complexity of design and debugging |
| Tightly coupled to where data is generated (IoT devices-local servers) | Low latency, local processing payload | Multiple sites to manage. |
The differention between grid computing and distributed computing is explained in one of the glossaries as follows: “Grid vs Distributed Computing: Grid computing tends to concern parallel processing for a single task, and distributed computing can be used to perform multiple tasks on separate devices.
Thin Clients and Network Computers
Another old, but still viable concept are network computers. Networked computer, or thin client, that are clients to a server and allow the user to access programs that reside on the server. The purpose is to decrease computing prices by simplifying the complexity of those of the customer’s machine.
A thin client is a low cost terminal which requires a lot of processing power from a server. Thin clients generally require a permanent network connection, and do little processing on the thin client itself, while management is on the server-side, often using a virtual desktop infrastructure (VDI). An even simpler version is a zero client. Thick clients are the other: they store most resources on their local machine and do not require continuous connection.
TechTarget credits Tim Negris of Oracle with coining the term thin client in 1993. A common feature mentioned is since the sensitive data is not stored on the thin client, it cannot be in jeopardy if lost or stolen.
According to ATIS glossary, this is known as a server-centric model, but suggests that most shopping Web pages are thin client because the client only has to install a browser and provide a network connection.
Cloud and Edge: The Modern Extremes
Cloud and edge computing are but two sides of the same coin.
- Cloud computing is the delivery of computing power and storage space on-demand remotely, particularly when it involves the Internet’s cloud resources. Large clouds distribute functions to several data centers.
- Edge Computing: Distributed paradigm, computation and storage near data sources. It is an architecture, not a particular technology and it dates back to the creation of content delivery networks (CDNs) in the late 1990s to deliver web content and video from content servers close to the user.
Many systems will actually employ both. One device could be used to work with critical information and the rest to the cloud for offline storage and analysis.
Where Does the Work Happen? A Four-Question Test
Work Happening Where-Test (Original Framework)
These four questions can be used to select a model. It is NOT an industry standard it is an editorial method.
- What is the tolerable delay in performing the task? Edge processing or local processing will be preferred if milliseconds are important.
- Where will data reside? Data, such as sensitive or regulated data, may be required to remain on a “controlled” server or on-site.
- What is the reliability of the connection? If the network is weak, then one can say that the direction is towards local processing – thin clients and cloud apps require continuous connection to the network.
- What should their cost do? Cloud-based subscriptions focus spending on usage, as owned servers and thin clients focus spending up front.
A hybrid design is often a practical solution, if two answers pull in opposing directions.
Design Checklist
This design checklist is a Linkable Asset that lists everything you should consider when designing a network.
- Recorded what the system will need to do and its clients
- Processed latency requirements (in milliseconds)
- Determined where to store sensitive data
- Verified some errors that occur on network failure
- Made the comparison of client-server and cloud/edge in terms of the workload
- Fixed Options: $770 Usage Based: $2600
- Planned security on all equipment to be connected.
- Thin clients, for locked down desktops in a managed environment
- Described the communication between components
- After 1 month of actual use, reviewed design
Guest Post Angles
These topics are good to put on networking, IT infrastructure, and cloud based publications:
- Thin clients return: What are VDI and secure desktops?
- Edge or cloud – which in which?
- Map of client-server, P2P, and distributed systems, for the beginning student.
Decide Where the Work Should Happen
Now it’s time to make decisions about where the work should take place.
Network Computing is really not one technology, but a collection of options regarding where processing, storage and control are found. Answer the four questions above about your own work before you choose a tool. Typically once I know where the work should occur, it becomes evident which model has been the appropriate one either based on client-server, cloud, edge-based or thin client model.
