As mentioned earlier, effective queuing techniques can be used to reduce the spread of delays.
Such a mechanism as a queue is used in any network device where packet switching is used, namely: in a router, a switch of a local or wide area network, a final node.
A necessity in the queue arises during periods of temporary overloads, when the network device does not have time to process incoming packets. The reason for the overloads may be the processing unit of the network device, in which case the unprocessed packets are temporarily placed in the input queue, i.e., in the queue on the input interface. In the case where the cause of the overload is the limited speed of the output interface (and it cannot exceed the speed of the supported protocol), then the packets are temporarily stored in the output queue.
There are several algorithms for processing queues:
- Traditional FIFO algorithm;
- Priority Queuing also called “overwhelming”;
- Weighted Queuing;
- Weighted Fair Queuing (WFQ).
Each algorithm was developed to solve certain tasks and therefore has different effects on the quality of service of various types of traffic in the network. A combined application of these algorithms is also possible.
The principle of the FIFO algorithm is the following: in case of overload, packets are placed in a queue, and if overload is eliminated or reduced, packets are sent to the output in the order they arrived (First In First Out). This queuing algorithm is applied by default to all packet-switched devices.
The mechanism of priority traffic processing consists in dividing all network traffic into a small number of classes. In this case, each class is assigned a certain numerical attribute – priority. The division into classes (classification) can be made in different ways.
The Weighted Queuing algorithm is designed to provide a minimum of bandwidth for all traffic classes or meet the delay requirements.
Weighted maintenance leads to longer delays and their variations than priority service for the highest priority class. However, for lower priority classes this ratio may not be fair, therefore, to create more favorable conditions for servicing all traffic classes, weighted maintenance is often more appropriate.
Weighted Fair Queuing (WFQ) is a combination that combines priority queuing with weighted queuing. There are various implementations of WFQ, which differ in the way of assigning weights and supporting different operating modes. The most common scheme provides for the existence of one special queue, which is served according to the priority scheme, that is, the first one, and until all the applications from it are selected. The remaining queues are viewed sequentially by the router according to the weighted maintenance algorithm.
Description Of Obtained And Planned Work Results
In the course of the work, the existing methods of providing quality indicators were analyzed and the one most suitable for further research was selected. This method is the network construction based on MPLS technology.
In the future, it is planned, to build a network based on MPLS technology to prove its advantage in practice using the OpNet modeling package. As for queuing algorithms, at this stage, developments are underway to improve the method of weighted fair service.
The most promising technologies for QoS are MPLS and Int-DiffServ because they combine the best aspects of both models. Thus, in MPLS, routing by labels allows for a more flexible allocation of network resources, which allows using several alternative routes of traffic delivery for creating high-speed backbones, consolidating local networks. In turn, Int-DiffServ represents the golden mean of the price / quality ratio. But for these technologies, there are restrictions that do not allow the use of these methods.
MPLS is still poorly distributed and is an expensive technology for corporate networks, and Int-DiffServ requires certain implementation costs and can not provide high compatibility rates.