What is a Network Model?

Oddly enough the Network model was designed to do what the Hierarchical model could not. Though both show how data is related the Network model allows for data to not only have many children but also many parents, where as the Hierarchical model allowed for only one parent with many children. With the Network model data relationships must be predefined.

It was in 1971 that the Conference on Data System Languages or CODASYL officially or formally defined the Network model. This is essentially how the CODASYL defined the Network model:

The central data modeling theory in the network model is the set theory. A set contains a holder record style, a set title, and an affiliate record type.

An affiliate record type is able to have the same role in more than one set; because of this the multi-parent hypothesis is established. A holder record style can be an affiliate or holder in another set as well.

The data model is an uncomplicated system, and link and connection record styles (often referred to as junction records) may well be existent, as well as additional sets between them.

Therefore, the entire network of relationships is demonstrated by a number of pair wise sets; within each set some record type is holder or owner (meaning one record type) this will be located at the tail of the network arrow (See figure below for an example) and one or more of the record types are presented as members or affiliates (the will be located at the head of the relationship arrow). Usually, a set defines a 1: M relationship, although 1:1 is permitted.

The Traditional Network Model

The most notable advantage of the Network model is that in comparison with the Hierarchical model it allows for a more natural avenue to modeling relationships between information. Though the model has been widely used it has failed to dominate the world of data modeling.

This is believed to be due to large companies choosing to continue using the Hierarchical model with some alterations to accommodate their individual needs and because it had been made almost obsolete by the Relational Model which offers a higher lever, and a more declarative interface.

For a while the performance benefits of the lower lever navigational interfaces used with the Hierarchical and Network models were well suited for most large applications.

Yet as hardware advanced and became faster the added productivity and flexibility of the newer models proved to be better equipped for the data needs.

Soon the Hierarchical and Network models were all but forgotten in relation to corporate enterprise usage.

{qbapagebreak title=OSI Network Model}

The OSI Network Model

Open System Interconnection or OSI models were created to serve as tools that could be used to describe the various hardware and software components that can be found in a network system.

Over the year we have learned that this is particularly useful for educational purposes, and in expressing the full details of the things that need to occur for a network application to be successful.

This particular model consists of seven separate layers, with the hardware placed at the very bottom, and the software located at the top.

The arrow identifies that a message originating in an application program in the column listed as #1 must make its way through all of the other layers contained in both of the computers in order to make it to the destination application in the column listed as #2.

This process could easily be compared to that of reading an email. Imagine Column #1 and #2 as computers when exploring the figure below:

The first layer, which is clear labeled as the physical layer, is used to describe components like that of internal voltage levels, it is also used to define the timing for the conduction of single fragments.

The next layer is the Data Link, which is the second layer that is listed in the example above, this often relates to the sending of a small amount of data, this could be and often is a byte, it is also often used for the task of error corrections.

The Network layer follows the Data Link layer, this defines how to transport the message through and within the network. If you can stop an moment and think of this layer as one working with an internet connection, it is easy to imagine that it would be used to add the correct network address.

Next we have the Transport layer, this layer is designed to divide small amounts of the data into smaller sets, or if needed it even severs to recombine them into a larger more complete set. The Transport layer also deals with data integrity; this process often involves a checksum.

Following the Transport layer we find the Session layer, this next layer is related to issues that go further or are more complicated then a single set of data.

More to the point the layer is meant to address resuming transmissions like those that have been prematurely interrupted or even some how corrupted by some kind of outside influence. This layer also often makes long term connections to other remote machines.

Following the Session layer is where we find the Presentation layer. This layer acts as an application interface so that syntax formats and codes are consistent with two networked or connected machines.

The Presentation layer Ialso designed to provide sub-routines as well, these are often what the user may call on to access their network functions, and perform some functions like encrypting data, or even compressing their data.

Finally we have the Application layer. This layer is where the actual user programs can be found. In a computer this could be as simple as a web browser surprisingly enough, or it could serve as a ladder logic program on a PLC.

Network Model Tips

After reading this article it is not hard to see the big differences between the Hierarchical Model and the Network Model. The network model is by far more complicated and deals with larger amounts of information that can be related in various and complicated ways.

This model is more useful due to the fact that the data can have many-to-many relationships, not restricting in to a single parent to a child structure. This is how the Hierarchical Model works with data.

Though the Network model has been officially replaced by the more accommodating Relational Model, for me it is not hard to imagine how it can still be used today, and may very well still be being used by PCs around the globe when I think of the Network Model in relation to how we email one another.

After reviewing the information and investigating the facts of the Network model I have come to the conclusion that it is a sound and relatively helpful model if not a bit complicated.

Its one major downfall being that the data must be predefined; this adds restrictions and is why a more suitable model was needed for more advanced data. Ultimately this one restriction lead to the model’s untimely replacement with in the world of data analysis.