QoS is an essential element of network architecture that is commonly used at the enterprise level. In environments where certain network functions can’t afford to be delayed or interrupted QoS should be implemented. A well designed plan with with QoS should increase the performance and efficiency of a network once established. This will help the organization’s overall performance by providing continuous access to business essential functions necessary for successful daily operations.
QoS uses four types of network traffic metrics to measure performance which are bandwidth, delay, loss and jitter. Each of these metrics provides a way to measure network traffic efficiency and where it may need to be improved. The first thing an organization should do is to identify those essential functions that use the most amount of bandwidth. This could be video teleconferencing, video streaming or VoIP services. In addition to bandwidth intensive activities there may be others that experience latency or packet loss. This can result from an overcrowded network which causes disruptions in service. This may also result from what is known as jitter which is when packets are received before or after their expected arrival time due to inconsistent transfer speed over the network (Fortinet, n.d.).
When these potential problems on the network are cross referenced with the most essential functions of the network the administrator can begin to prioritize certain types of traffic. From here the network administrator can determine what methods are best for guaranteeing the best performance and efficiency on the network. One method of accomplishing this is a technique known as traffic marking. This is where specific types of traffic from essential applications is marked on the network. From here it is then configured to take priority over other applications on the network that may also need bandwidth. This is a very straight forward method of QoS (Microsoft, 2016).
Another technique that could be used is known as queuing. To use this method the network administrator will first set up a dedicated queue that will buffer resources that will later be used on the network for a specific task. From here the important traffic type will be identified and the network will be configured to reserve this queued bandwidth for that specific service. This method is slightly more complex but accomplishes the same thing as traffic marking. However, in this case the network automatically guarantees that the application will have access to the resources it needs by reserving it ahead of time (Fortinet, n.d.).
For example imagine that there is a call center with hundreds of employees. Each of them is using VoIP services to contact customers, some are using video on demand (VOD) for training and others are using chat services and a variety of other applications. The traffic on this network could become very complex and without QoS mechanisms in place the network availability needed for mission business tasks could be disrupted. In this instance it would make sense for the network administrator to configure a resource reservation queue for VoIP traffic since contacting customers is the primary objective of the call center. By creating this policy the employees would always be access to the VoIP service despite it being a resource intensive service (Microsoft, 2016).
In an enterprise network there are thousands of information systems competing for network resources. If these devices are not subject to some sort of QoS strategy then there is a good chance that the most critical functions may not be available when they are needed. There are several other methods of configuring QoS on a network such as by port number, type of service or application, resource reservation and several other internet protocols. Understanding the network and the services that use it is key to determining how to use QoS for increasing performance and efficiency. Although there are many best practices there is no one size fits all approach to QoS. Each network administrator will need to identify inelastic services that are more susceptible to network disruptions listed above.