Estudio comparativo de algoritmos de colas en calidad de servicio de redes utilizados en mikrotik

Abstract

Considering the aforementioned, the present research work consists of carrying out a comparative study of queuing algorithms used in mikrotik, of which we have chosen 4. RED, CODEL, FIFO and CAKE, and through a network topology in the GNS-3 simulator , perform tests with data flows. The project has several phases that go from a bibliographical research to the tests in the simulator. In the first point, the identification of the problem, the need to place the Qos algorithms in the networks, the search for data begins taking into account the most relevant data for the investigative work. Then the network topology that is going to be simulated in GNS-3, as a third point, a table of characteristics of the algorithms with data chosen from the bibliographical research that was carried out in books, magazines, internet sites, analyzing the information through comprehensive reading, and with an external program called Flent used in the server with Iperf that sends data flows between client and server, and to visualize the results in its virtual environment, measure through QoS metrics (latency, ping, bandwidth (upload and download) and packet loss rate to understand the performance and performance with experimental tests, then scale the algorithm ratings from 1 as the lowest to 5 as the highest. that were obtained through the tests, they show us images and textual evaluations with averages, indicating that, during the execution of tests, the algorithms had differences in some parameters. The interpretation of the data, the differences were reflected in the latency and stability of each one, Cakebeing the best valued, it obtained 1.77 ms of average PING in download and 1.88 ms in upload, and with a stability in sending and receiving of packets as well as the CoDel algorithm that had high pings in the upload and download of an average of 19.43 ms and 19.50 ms, the FIFO algorithms 88.81 ms, 88.90 ms respectively and RED 30.72, 30.96 were more unstable at the moment of the test results. And finally, recommendations on future improvements with the studies carried out based on the network service quality project.