QoS Fundamentals

Overall rating: 3.78 Instructor: 4.25 Materials: 3.22 more …

In this webinar, Ethan Banks will discuss what Quality of Service (QoS) is for, and when and where it should be used to solve specific business problems. Ethan’s perspective is from that of rolling out QoS schemes across a number of globally distributed WANs. In that setting, he’ll explain fundamental QoS technologies. You’ll come away with foundational knowledge about:

• Packet marking schemes
• Traffic classification
• Trust boundaries
• Congestion avoidance
• Congestion management
• Queue depth
• Tail drop
• Bufferbloat
• Modern active queue management techniques such as CoDeL

While all of that sounds technically heavy (and it is), the content will be delivered in a friendly style that provides context around QoS tools. You’ll always know why we’re talking about what we’re talking about so that you can sort QoS tools into the proper drawers in your toolbox. With proper application of the information, you should be able to begin forming a QoS plan for your organization.

Contents

The webinar will be delivered in a single session over an expected runtime of 2-3 hours, depending on audience questions.

What is QoS good for?

In this introductory section, Ethan will discuss a real world perspective on quality of service. What does it actually do? What can’t it do? When is it the right set of tools to apply to a networking problem, and when is it not? He’ll also address misconceptions about QoS.

My traffic is more important than your traffic

In this section, Ethan will cover how to differentiate traffic flows from each other. How does a network device distinguish between a voice flow and an HTTP flow? And how is that classification preserved throughout the network infrastructure? What happens to classification schemes when traffic hits a boundary such as the Internet or WAN provider? To answer that question, Ethan will discuss DSCP marking schemes, realistic traffic classes, trust boundaries, and mutation maps.

The ocean and the straw

In this section, Ethan will describe what happens when a link is congested. That is, what is going on inside of a network device when there are more packets to send than the interface is able to deliver because it simply isn’t fast enough? When a bunch of packets are queued, which ones get sent out first? Which ones get dropped? To answer those questions, Ethan will discuss low latency queueing, class-based weighted fair queueing, traffic shaping, policing, and weighted random early detection.

Can’t we just let TCP be TCP?

In this concluding section, Ethan will discuss the problem of bufferbloat, explaining the problems that it causes for TCP, including an increase in congestion. He’ll then review the modern active queue management technique CoDeL as an example of how congestion avoidance is being thought about in the modern era. Ethan will also point out modern TCP implementations that are re-thinking congestion avoidance such as CUBIC. Ethan will also introduce Google’s emerging BBR algorithm.

The Author

Prerequisite Knowledge

The ideal engineer for this webinar will understand IP packet formats and the basics of TCP flow control, such as acknowledgements, windowing, and the slow start algorithm, although this is not required to benefit from the discussion. There is no assumption of certification level or vendor alignment. The content as delivered will be vendor agnostic.

Target Audience

• Networkers new to QoS looking to ramp up quickly.
• People intimidated by QoS who want to dispel FUD about the topic.
• Engineers with a specific network problem they believe QoS might solve.
• Architects who wish to review of QoS fundamentals.