Standards, ISO-OSI

Standards are an efficient way to keep the same interface of different elements in a large set. Some of them are imposed by manufacturers, some others are commited by international organizations. Some of them are born in IT field and others come from telecommunications field.

ISO-OSI is the international standards organization — open systems interconnection. It defines a layered stack of integrated models for the different implementations of (not only) internet communications.

The ISO-OSI stack

The model is a standard architecture, both hardware and software, layered. Each layer provides some services to the upcoming layer. The highest the layer, the higher is the abstraction level. Each layer matches an abstraction level, and each layer provides more defined functions than the previous layer. The data exchange between levels is minimised.

Each level adds a header containing additional informations able to allow the packets to be forwarded between each level. The data created by an n-level protocol are called n-pdu (n-protocol data unit).

A pdu consists of a header and the actual payload.

Physical layer: it interfaces directly with the physical medium, abstracting it: it offers a medium indipendent transmission to the second layer.

Offered services:

  • physical link management between two adjacent systems
  • transmission of each pdu bit
  • bit sequency deliver
  • failure notification

Data link layer:

  • faces failures of the physical layer (error detection and potential error correction).

Offered services:

  • pdu transfer between adjacent systems
  • selection of a certain service quality (when possible, based on level 1 frequency and bandwidth capacities)
  • usage of two packet send and receipt queues

Net layer:

Knows the net topology and routes the packets. If a connected service is offered, it establishes, manages and releases net connections.The service allows information transfer between each end, allows a certain quality wherever possible (with such parameters as latency) and switching.

Connected services and protocol, unconnected services and protocols:

If a service is connected, there are primitive structures available to show up to the interlocutor, such as establishing a connection, for messages exchange, and finally break down the connection. If a service is not connected, there are only available primitive structures to send messages, sent separately. Connected protocols are used to create connected services.

Transport layer:

Fills in for weaknesses and quality fluctuations of the net layer service. It’s the first end-to-end layer. Offered services include: connection establishment, data transfer, connection management, connection release, synchronisation between the two systems by confirm send and requests.

Techniqus used in level 2–3–4 protocols:

  • feedback system by acknowledgement
  • piggybacking (maximum wait time)
  • window controll techniques

Session layer: Synchronizes the dialogue between the two processes.

Presentation layer: Allows to exchange messages regardless of the transmission syntax.

Application layer: It’s the mean of accessing to the net for an application process.

Session, presentation and application layer are more often in series than in parallel, and are often merged.

A connection can be implemented at n-level using more n+1 connections, or a n+1 connection can use more n-level connections.

The stack supports constant over time connections at n+1-level but inconsistent at n-level, or constant over time n-level connections and inconsistent at n+1-level

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.