Quality of Service

from Wikipedia, the free encyclopedia
This article is not adequately provided with supporting documents (for example individual proofs ). Information without sufficient evidence could be removed soon. Please help Wikipedia by researching the information and including good evidence.
It is unclear from which books or other major sources the information in this article was drawn .-- Stegosaurus Rex 11:35 am, Feb 27, 2012 (CET)

Quality of Service ( QoS ) and quality of service is the quality of a communication service from the user's perspective. That is, how closely the quality of the service matches their requirements. Formally, QoS is a set of quality requirements for the common behavior or interaction of several objects.

conditions

Standard IEEE 802.1p

  • A user wants to be reliably connected to the desired destination and be reliably disconnected after communication has ended.
  • The connection should be established quickly.
  • Problems when establishing a connection (e.g. target station cannot be reached) should be reported to the user as quickly as possible.
  • A communication connection should remain stable.
  • The communication participants want to be able to understand each other.
  • The information should be transmitted completely and without errors.
  • No information from other communication participants and no faults should be transmitted.
  • Communication should be as true to the original as possible.
  • There should be no long waiting times during communication.
  • The billing of the communication should correspond to the correct amount of time and data.

In order to be able to measure such general user requirements, quality parameters are defined that are intended to describe these requirements.

  • Quality of the connection establishment: percentage of connections that were not established, percentage of incorrect connections, percentage of double connections, percentage of slow connections.
  • Quality of an existing connection: percentage of premature connection interruptions, percentage of information not transmitted, percentage of incorrectly transmitted information, delays in information transmission, effective bandwidth of the transmitted information, fluctuations in the signal delay, proportion of interference in the transmitted signals, echo components.

By measuring these network-related parameters and assigning them to user requirements, one tries to quantify the quality of the service.

Since the provision of a service with high quality requirements is correspondingly complex, network operators are switching to charging high quality standards with higher prices. Here, compliance with certain quality parameters is promised for a certain quality standard. The customer can request a certain Quality of Service (see Service Level Agreement ).

Compliance with high quality standards requires the proper functioning of all interacting components of a telecommunications network . The operation of the technology must be continuously monitored, error parameters must be recorded and recorded and thus form the basis for any maintenance measures that may be required.

QoS in telecommunication networks

Essential categories of QoS in communication networks are the traffic quality, which deals with the switching technology, and the transmission quality , which deals with the transmission technology.

In the UMTS mobile communications, four QoS classes, for example, been defined:

background
for data transfer with the lowest possible error rate but uncritical requirements for bandwidth, delay and jitter.
conversational
for direct communication (telephony, video telephony). Requirements similar to streaming, but significantly less jitter and delay are permitted.
interactive
for the use of interactive services. Similar requirements as with Background, but higher requirements for the delay in order to avoid waiting times when using the services.
streaming
for distribution services. A minimum bandwidth is required; Jitter is permitted to a certain extent, since jitter buffers are used on the receiver side. Bit errors are rather uncritical.

QoS in switching technology

The grade of service refers to the quantifiable part of the QoS that is dependent on the respective rating of the equipment. The traffic quality of a communication network depends on the number of lines and control devices. In the telephone network , the switching centers are the control devices, in the Internet they are the routers . Quantitative descriptions of the traffic quality use parameters of traffic theory such as waiting probability , average waiting time or loss probability . In switching technology z. B. include the following influencing variables in the QoS:

  • Double connection: a connection in which one or more additional connections are connected to the desired connection
  • Incorrect connection: a connection in which a subscriber other than the desired subscriber can be reached despite the correctly entered number
  • No release: A connection is maintained, although "Release" was requested
  • Premature disconnection: Disconnection of a connection without being initiated from the calling or called connection
  • Counting fault: incorrect recording of the charge-relevant parameters of a connection

In this example, the measured and weighted counts of such events then form a component of the traffic quality of a specific exchange .

Other factors influencing the quality of traffic are:

  • Blocking: the state of a network in which the establishment of a connection cannot be completed because the necessary technical facilities or resources are not available.
  • Delay in sending: Time that elapses between the start of input at the origin and the start of output at the destination
  • Delay in connection control signals: Example: The time between the signals "outgoing call" ( telephone receiver hung up) and "dial request" (dial tone) is too long.

QoS in IP networks

The quality of service of services that are transmitted over IP networks - networks that use the Internet Protocol - is influenced by additional IP-specific parameters. If IP is used as the transmission protocol in a telecommunications network, it forms a higher layer in this network according to the OSI model . In this case, both the transmission quality of the lower layers and the transmission parameters of the IP level are relevant for the assessment of the overall quality of service of a service based on IP technology. The quality of service perceived by the user results from the interaction of all quality impairments. If, for example, it is an Internet connection that is dialed in via a public telecommunications network, the entire quality of service is formed from the transmission, traffic and switching quality of the transmission channel via the public telecommunications network and the quality of the higher IP layers.
If IP is not transmitted over a public telecommunications network but, for example, in a LAN based on Ethernet , only the impairment of quality of this LAN and those of the IP transmission have to be taken into account.
A particular factor in the quality of service of services that are implemented with the help of the Internet Protocol is that, in contrast to services in traditional telecommunication networks, it is significantly influenced by the terminal. In conventional (circuit-switching) telecommunications networks, on the other hand, the transmission, switching and traffic quality was mostly related to the quality of the network; the influences of the end devices could be neglected.

In IP networks, the influence on QoS is usually recorded using the following parameters:

  • Latency : the delay in end-to-end transmission
  • Jitter : the deviation of the latency time from its mean
  • Packet loss rate : the probability that individual IP packets will be lost during transmission (or - with real-time services - reach their destination too late)
  • Throughput: the average amount of data transferred per unit of time

Different services have different requirements for these parameters. The latency is particularly noticeable in applications that require short response times ( e.g. Telnet ). For example, in the case of high latency, an input only appears on the screen with a certain time delay. For pure file transfers, the overall throughput is usually the decisive parameter, but the individual latency and loss rate are less important here. For real-time communication such as B. Voice over IP, on the other hand, latency, jitter and the rate of loss play a far greater role because they have a decisive influence on speech intelligibility. As a real-time application, IPTV even has very significant demands on the overall quality of service, as even small quality defects in the transmission have a visible effect on the image display on the television. The throughput is important because videos often have a high data rate required and failure to provide this rate the video is simply stopped.

Within a transmission channel, it may be necessary to increase the QoS for certain data streams at the expense of other data streams. This can be done, for example, by prioritizing IP data packets based on certain characteristics and properties. With these mechanisms it is possible to give preference to certain services such as Voice-over-IP , which requires a constant data stream with constant delays, than downloading from a file server (FTP) or calling up websites.

Certain reservation protocols in the network are used to ensure that the isochronous data stream can be guaranteed for the entire duration of a data communication .

IPv4 and IPv6 packets have a flag ( DSCP Differentiated Service CodePoint ( RFC 2474 ); formerly Precedence ( RFC 791 )) in the IP header , which identifies the type of data in this packet ( traffic class ). Based on this flag, the data packets are prioritized (i.e. handled preferentially). Describes the current considerations ( RFC 3260 ). There are a number of other methods of managing quality of service.

Realization in IP networks

QoS on layer 3

On the theoretical level, QoS can be implemented through prioritization or parameterization of the data traffic, data rate reservation, data rate limitation and packet optimization. From a technical point of view, there are two mechanisms for this:

  • Either you register pending data flows with all active network components (router, etc.) and reserve the required data rate ( IntServ , Integrated Services),
  • or all data packets are marked and the active network components treat / prefer the packets according to their markings ( DiffServ , Differentiated Services).

Since IntServ, in practice mostly using Resource Reservation Protocol (RSVP), results in a high management overhead and even a device that does not support IntServ causes the whole mechanism to fail, the alternative DiffServ has meanwhile established itself. DiffServ is also more scalable. The term Type of Service (ToS) is also often used. 1 byte was reserved in the IP header for ToS, of which only 6 bits were used by ToS. However, no binding standard could be established that all manufacturers of network equipment adhere to. In the meantime, the six most significant bits of the former ToS field in the IPv4 header and the 'Type of Service' octet in the IPv6 header have been redefined by the Internet Engineering Task Force (IETF). They are now uniformly called Differentiated Services Field and can be evaluated to a numerical value Differentiated Services Code Point (DSCP).

Since the practical implementation of extensive QoS measures on the Internet has hardly developed further for years, there are critical voices as to whether QoS mechanisms on the Internet are even necessary or suitable.

One reason for the stagnation is the fact that DiffServ has not yet been able to establish explicit numerical values ​​for QoS parameters such as jitter and packet loss rate, nor even a uniform use of the code points and a uniform definition for the per-hop behavior. The EU project MUSE of the 6th framework program of the EU remedied this deficiency by defining four QoS classes for which these values ​​are fixed. In another European research project PLANETS, a practical implementation proposal was derived from this, in which even the lengths of the queues are specified.

QoS in ATM networks

When establishing a connection in ATM networks, the end system negotiates with the service provider a service class / QoS class that meets its requirements, which is then guaranteed for the connection:

  • Constant Bit Rate (CBR) : Transmission at a constant data transmission rate .
  • Variable Bit Rate (VBR) : The data transmission speed can change during transmission. With VBR one also distinguishes between:
    • VBR-rt : The transmission has real-time requirements (rt = real time).
    • VBR-nrt : The transmission has no real-time requirements (nrt = non real time).
  • Available Bit Rate (ABR) : The transmission should always take place at the currently available data transmission speed.
  • Unspecified Bit Rate (UBR) : As ABR, but with UBR no QoS parameters are defined.

QoS parameters:

  • Cell Error Rate (CER) : Ratio between incorrectly received ATM cells and the total number of received ATM cells.
  • Delay (Cell Transfer Delay - CTD) : time required by a cell from sender to receiver.
  • Jitter (Cell Delay Variation - CDV) : Fluctuations in the transit time of ATM cells in a connection through the ATM network.
  • Bit error rate (Bit Error Rate - BER) relationship between erroneous bits and the total number of received bits.
  • Cell Loss Ratio (CLR) : Ratio between lost ATM cells and the total number of received ATM cells, see packet loss rate .
  • Cell Misinsertion Rate - CMR : Ratio between misdirected ATM cells received and the total number of ATM cells received.
  • Severely Errored Cell Block Ratio - SECBR : Ratio between incorrectly received ATM cells and the total number of an ATM cell block.

The best known queuing mechanisms are:

  • FIFO (First In - First Out) - The data packets that arrive first are also forwarded first
  • LIFO (Last In - First Out) - The data packets that arrive last are forwarded first
  • LLC (Low Latency Custom-Queuing) - A short runtime of previously defined packages is guaranteed
  • Priority queuing - the data packet with the highest priority is forwarded before the lower priority packets.

For details on individual queuing algorithms, see Network Congestion Avoidance

Individual evidence

  1. ↑ cf. Bandwidth vs. QoS ( Memento from July 29, 2013 in the Internet Archive ) and Why We Don't Need QOS: Trains, Cars, and Internet Quality of Service
  2. ↑ http://www.medea-planets.eu/QoSsolution.php?page=solution
  3. ↑ Gerd Siegmund: ATM - The technology. 4th edition, Hüthig, Heidelberg 2003, p. 276
  4. ↑ Gerd Siegmund: ATM - The technology. 4th edition, Hüthig, Heidelberg 2003, p. 272

Web links