Quality of Service, Part 1


Quality of Service (QoS) is a common subject in the classroom. There seems to be an element of trepidation from students with regards to their understanding and comprehension of QoS. In considering important topics to address, QoS stood out as a subject most students want to talk about.

While reviewing the available reference materials, the reason behind the trepidation I had sensed from some of my students is clear. The books and white papers on QoS are lengthy and packed with exceedingly complicated subjects. So, I have decided to write a blog series to help simplify the mysteries of QoS. Let’s get started.

QoS ensures more predictable network services by providing dedicated bandwidth, controlled jitter and latency, and improved loss characteristics. QoS provides tools for managing network congestion, shaping network traffic, using WAN links more efficiently, and setting traffic policies across the network. QoS helps provide consistent, predictable network performance by offering intelligent network services.

With all of the above being said, it is no wonder so many students feel uneasy with the subject of QoS. In this series of blogs I will take a closer look at the many functions of QoS.

Where it all started
In any review of the public switched telephone network (PSTN), it is important to note it was all built with circuit-switched networks. These networks were fixed bandwidth networks with dedicated circuits that were well suited for real-time traffic, such as voice. In addition, most enterprises had multiple networks, where data was carried on one network and voice was carried on another. Multiple networks allowed for dedicated network resources for specific traffic flows. Usage of the PSTN for internal calls and maintaining multiple networks was extremely costly and labor intensive for most companies.

With the popularity of Internet Protocol (IP) and packet-switched networks as the underlying fabric of the internet, the ever-shifting paradigm for networks became “everything over IP”. There are a multitude of applications such as Voice over IP (VoIP), streaming video, e-mail, e-commerce to name a few. The underlying issue was that IP was designed to provide best-effort service for delivery of data packets, where all network packets are treated exactly the same with no preference given to time sensitive packets. Different applications have varying needs for access to the network resources and how their packets are handled on the network.

Another popular term in the industry is “converged network” meaning to combine voice, video, and data on a packet-switched network running IP. The one technology that enables IP to converge all these packets is QoS. In essence, QoS is the differential treatment of the voice, video, and data packets that flow on the IP network, creating a system of managed unfairness. QoS technologies allow different types of traffic to contend inequitably for network resources. Time-sensitive applications, such as voice or interactive video packets, can be given priority over data applications.

With converged networks merging many different traffic streams, each with different requirements, problems can arise. Voice traffic is typically small and cannot tolerate delay as it traverses the network. Data packets, such as file transfers, are typically large and can survive delays and drops with retransmissions of the packets. Small (voice) packets compete with bursty data flow on the converged network, with QoS acting as the mediator to ensure priority goes to time sensitive packets.

For the network to provide secure, predictable, measurable, and sometimes guaranteed services, the fixed qualities of a network and the flow of packets must be managed with QoS. Some of the issues that can occur within a network that can have an impact on our time-sensitive packets are:

  • Bandwidth – Lack of bandwidth on the network the IP packets are traversing.
  • Packet Loss – Dropping of packets because of network congestion, not network outages.
  • Delay Variation (Jitter) – The time difference between how long it takes packets to traverse the network.
  • Out-of-Order Delivery – Different packets may take different routes and arrive at the destination in a different order than they were sent.
  • Delay – The time it takes to get the packet end-to-end, or from the mouth to the ear.
    • Packetization Delay – Time required to sample and encode voice or video into an IP packet
    • Serialization Delay – Time required to put the packet on to the wire
    • Propagation Delay – Time required for the packet to traverse the media

A few of the ways we reduce the effects of limited bandwidth and delay on the network are as follows:

  • Upgrade the links to increase bandwidth, though this can be an inefficient use of resources and very expensive
  • Compress the payload of the packet (voice or video)
  • Compress the IP header (CRTP)
  • Forward the important packets first

Forwarding the important or time-sensitive packet first, or differential treatment of packets, will be the focus of the next few posts. In subsequent posts the different types of QoS will be examined, as well as how they operate, and the tools used to implement QoS.


Author: Paul Stryer

In this article

Join the Conversation