Quality Of Service (QoS)
In the field of telecommunications networks, the term quality of service or simply QoS (Quality of Service from English) is used to indicate the parameters used to characterize the quality of service provided by the network (e.g., packet loss, delay) or the means to achieve a quality of service desired.
The quality of service is usually negatively correlated with the traffic offered to the network, and positively with the resources committed to implementing and managing the network.
The traffic offered to the network and the involvements of failures are random processes, hence the parameters used to characterize the quality of the service are random variables.
When a service contract provides the parameters of quality of service, with penalties if these parameters are not met, we speak of SLA or SLA (service level agreement).
In the field of telephony and in general the circuit quality of service includes parameters such as:
- Noise level on the circuit
- Noise level
- Probability of finding a free line to start a communication
- Likelihood of unwanted interruption of communication
- Service availability
- Average duration of disruption and maximum
In a packet network, a packet received by a switch can find the port that should be undertaken by another broadcast packet transmission. In this case, it is stored in a queue, and suffers for this delay. If the queue is full, the packet is discarded.
The parameters typically considered for a packet network are:
- Loss of packets or dropped packets: it is considered the percentage of packets that the network as a whole fails to deliver to destination. The loss of a packet is handled differently by transport protocols, although this falls outside the definition of quality of network service: a protocol without feedback, it was the failure to disclose information, as reflected in a protocol with TCP, the receiver, after waiting a reasonable time, should request that information be retransmitted, causing serious delays (delay) in the overall transmission.
- Delay (delay) incurred by a packet entering the network from its delivery to the recipient. Characteristics are considered as the average delay (packets on average takes 10 ms to traverse the network), and its percentile (99% of packages are delivered within 20 ms). Jitter is also considered, namely the delay variation between packets sent in sequence from one node to another.
- Out of order delivery, or out-of-order – on some networks, it is possible that a sequence of packets sent from one node to another are delivered in a different order than the original.
- Transmission error, a packet can be delivered to, but not identical to that sent. Many networks recognize the majority of transmission errors, and some are even able to correct these errors. It considers the percentage of erroneous packets. Normally, transport protocols recognize a bad packet and the retransmission request as if it had been lost, but it is also possible that the error reaches the end application.
Applications requiring QoS
The original Internet model of QoS or no QoS, it is suitable for elastic applications that can run on networks with very degraded performance, and vice versa use all the bandwidth available when it is abundant.
Other types of service are called inelastic, or require a certain level of bandwidth to function – it does not get more leverage and they get less will not work at all. These are applications that require the adoption of measures to ensure a certain QoS.
Applications that require QoS such as the following:
- Multimedia streaming: may require a guaranteed throughput
- VoIP may require very tight constraints on delay
- Dedicated link emulation requires both guaranteed throughput limited to a maximum delay
- Safety critical application such as remote surgery may require a guaranteed level of availability; this is also called hard QoS.
In the workplace, it is possible for the definition of QoS requirements for applications that are not inherently flexible, to ensure adequate levels of productivity. For example, the “end of the travel agency must be able to complete the transaction within 10 seconds in 98% of cases.” But often a requirement of this type of action is on the network information system that provides the service (for example, set an appropriate number of servers).