Quality of Service (QoS) is the “description or measurement of the overall performance of a service, such as a telephony or computer network.” Networks run many different services and applications, so the goal for each service or application is to establish an optimal way of running it that assigns it the network priority and resources required to run it in a mix of network traffic.
To do this in an ideal world, a quality-of-service engineer looks at bandwidth for maximum rate of transfer, throughput for actual rate of transfer, latency for the amount of time it takes a packet to move across the network from start to finish, and jitter, which occurs when packets arrive in a different order than they were sent.
Applications like video conferencing and voice over IP (VoIP) can be highly sensitive to jitter or latency, so these applications must be routed over the network in ways that minimize latency and jitter. In other cases, such as the transport of a large batch file, reserving the necessary bandwidth is the key.
The bottom line is that all network services and applications have their own unique needs. Collectively, all of these different needs must coexist and run optimally over the network.
Server and network operating systems and utilities enable network specialists to set priorities for running different applications and services. Unfortunately, these utilities often aren’t granular enough for fine-tuning. In some cases, such as in a heterogeneous computing and networking environment, these tools might even conflict with each other.
Approaching QoS Holistically
Network solutions providers recognize the quandary of running a mix of different services with each service having its own management tools. This is why some providers approach the management of these different services by offering an over-arching, single toolset that looks at everything running over the network and how each service and application has to run so it can deliver the best performance.
Among vendors in the end-to-end network QoS space are Fortinet, Solarwinds, ManageEngine, and others, as well as Opensource providers like OpenNMS. The focus is on providing a single network management framework that can address QoS while also reducing the number of different tools from assorted vendors that network analysts use. The tradeoff for sites using a single, overarching set of tools is that they might sacrifice some network management autonomy and risk vendor lock-in by going for an all-in-one solution. The other caveat for QoS is that software alone won’t bring QoS into your network. QoS begins with policy setting, categorizing different network services, and defining runtime metrics. These tasks are the job of network professionals.
Digging down into the QoS weeds
When network administrators adopt QoS, it often happens because of an intervening business event that dramatically changes how the company must use its networks.
In one case, a medical center in Nevada had to find ways to connect with patients in outlying rural areas who could not make it to the Center for in-person services. The solution was telemedicine and teleconferencing, but the company's network group had never planned for prioritizing video as a mission-critical delivery service. This sea change in business required the networking group to rethink how they had set priorities for network traffic and to formulate new policies on traffic and bandwidth priority usage.
Policy-making and priority setting for the network became a business issue instead of a background task that the networking group tackled on its own. When telemedicine and teleconferencing became mission-critical service delivery channels, leaders on the business side had to sit down with IT to redefine company and network priorities. For example, if telemedicine took priority over all other types of network traffic, other services, such as file transfers or email routings, had to be subjugated to this mission-critical workflow—and a consensus of this change had to be obtained from users and management.
Steps to implementing QoS
Those actions are the first step in QoS work. Namely, policy setting.
The next step is assigning metrics that measure performance.
For example, in the case of a telemedicine call, you don’t want the call to drop, nor do you want video frames to freeze or blur. To achieve and maintain the metrics for the level of service, QoS for routers and ports along the video route, as well as for other network assets, must be defined. If maintaining this metric requires slowing or halting other network traffic so that adequate bandwidth and throughput for video calls are maintained, that will be the step that IT takes. The consequence of this network service prioritization should be explained to end users so that everyone understands that if a situation emerges where network traffic becomes congested, other services may be delayed.
Once performance metrics are assigned, network assets must be properly configured.
To implement QoS on very large networks, individual ports and pieces of equipment need to be reviewed for their traffic settings so that every asset along the path of a particular network workflow is configured to achieve that workflow's performance metrics. Often, this is a painstaking task that used to be done "by hand," and many sites choose to work out with an outside QoS specialist.
Finally, network QoS must be continuously monitored, maintained, and fine-tuned when needed.
There is a plethora of network vendors that offer tools for monitoring network QoS. A methodology and a toolset need to be selected so that the QoS policy and performance metrics that have been established are upheld.
A final word on QoS for modern business apps
In the past, and still in many companies today, IT sets network priorities for applications and services based upon general IT best practices, and in many cases, this is all that is needed.
However, as more organizations bring on mission-critical applications and services like teleconferencing, real-time work collaboration, Internet of Things (IoT), AI, and robotics, simply abiding by traditional IT best practices for network performance may not be enough.
This is where QoS and a conscious setting of business priorities in the network begin.
By implementing QoS, a company can measure and optimize the performance of mission-critical applications and services, while other services and applications that are of less importance can run at lower priorities. There must be user and management consensus on this QoS priority setting before QoS is installed.
Gaining user understanding and then installing QoS onto networks are what help organizations comprehend how their networks are running and the value that networks are delivering to the business. These are also ways to ensure that network priorities stay in step with the business.