Network quality of service is the measurement of the overall performance of a network, and the purpose of QoS is to have the network performing optimally for the applications it must support.
Everyday network communications and traffic management don’t demand the same high levels of quality of service and bandwidth as, say, a telehealth meeting between a doctor and patient, nor does it require the same network management that thousands of edge IoT devices, sensors and equipment do.
In the case of extremely high-bandwidth, mission-critical applications like these, the network staff must be able to guarantee optimal quality of service at all times that these applications operate. This means analyzing and tuning every element of the network, whether it’s measuring and limiting packet loss, monitoring bit rates and data throughput, taking steps to ensure that the network provides continuous uptime and availability, or uncovering and solving any element of delay that could slow network performance.
SEE: Don’t curb your enthusiasm: Trends and challenges in edge computing (TechRepublic)
The QoS quandary
The problem is that most IT network staff lack the time, and in many cases expertise, to develop a QoS network.
The legacy applications of networks were apps that carried a system of records data only, so for many network professionals, it was sufficient to deploy applications and adjust the network that ran them in a straightforward way that only had to consider whether the throughput of data had enough bandwidth. If it didn’t, more bandwidth could be allocated — or perhaps a run schedule for the application could be worked out so the app only ran at non-peak times of throughput demand.
Now networks must carry not only a traditional system of records data, but also large files of unstructured big data and applications such as live video conferencing. With the introduction of IoT in the field and in manufacturing, the incoming streams of live data have also grown exponentially.
The rule of thumb is that data can only travel as fast as its lowest common denominator pipelines and connection points can carry it. If you’re using a 5G gateway for your network portal, 5G can deliver up to 20 Gigabits-per-second peak data rates and 100+ Megabits-per-second average data rates, but your data stream must also pass through a connecting router on your network that tops out at 1300 MBps.
The router is the slowest cog in the data line. That router will slow your entire data transmission. You might even find that some routers in your network won’t even run with 5G. What then?
Getting to QoS
“We had a dilemma like this,” a network manager for a west coast health clinic told me. “We needed QoS to support the telehealth applications of our physicians and our patients, but the network was never set up to provide the level of service that a continuous, high quality video conference of 20-40 minutes would require. We started to analyze our entire network for performance, but soon found that our normal day-to-day network tuning skills could not accomplish the level of QoS performance that would be needed. We determined that we had to hire a consultant who specialized in QoS.”
The QoS consultant was able to do several things. First, they reviewed and updated the documentation of this clinic’s network topology, router configurations and levels of software on each device on the network. This was a task that internal network staff could do, but there never was the time to do it.
From this work, what staff discovered was that their network documentation was out of date. They really didn’t have an accurate understanding of all the routers and devices on the network, nor did they know the levels of software on each of the network devices. In some cases, it was necessary to upgrade software to the latest level for optimal performance. In other cases, it was necessary to rip and replace devices.
Second, once network documentation was updated, the QoS consultant sat down with network staff to determine the QoS standards of performance that the network would need to perform in order to deliver high quality, uninterrupted video conferencing.
Once QoS performance requirements were defined, the consultant then looked at the CPU utilization of each router at the times when each router was busiest. Slow network links were also identified during this process. Next, bandwidth requirements were analyzed so bandwidth on the network could be re-calibrated to support the QoS requirements for video conferencing.
Finally, QoS performance parameters and priorities were laid out for each application on the network, with QoS policies documented and then implemented for the mission-critical video conferencing. The clinic was now in a position to move forward with QoS for its telehealth app.
“It’s expensive to hire a QoS consultant, but what we learned in the process was that we were not well-prepared in our own right to implement the level of QoS that we needed for telehealth,” said the network manager.
As more companies move mission-critical applications like telehealth and smart manufacturing to the edge, QoS requirements will grow.
Many sites will find themselves in the same position that this clinic did: Lacking the time and skills to deliver the level of QoS that is needed.
Even if this happens and you must hire a consultant, there are still steps network professionals can take to assert more control over their network QoS in the future.
These steps comprise the following:
- Always keeping network documentation up to date.
- Always installing the latest levels of firmware and software on all network routers and devices — and, of course, on the network itself.
- Ripping and replacing routers and devices that are incompatible with a new broadband technology such as 5G, and allocating budget to support these updates.
- Engaging in daily conversations with end users on future needs and applications that will require new and higher levels of network QoS so the future can be planned for.
Read more about edge computing with a look at its history, the benefits and how it will support the metaverse.