Automatic Access Point Capacity Management: Preseem Webinar

Automatic AP Capacity Management webinar recap

We recently held a webinar to celebrate the launch of Automatic Access Point Capacity Management in Preseem, as well as to provide an overview of our approach to quality of experience (QoE). The webinar was hosted by Dan Siemon, Chief Product Officer, and Jeremy Austin, Senior Product Manager at Preseem. Read on for a recap of the event or watch the full webinar here.

Our QoE Definition

QoE is 80% having the tools to proactively address issues before they affect customers, by making sure the network is operating cleanly and has sufficient capacity. The other 20% is mitigating issues that might otherwise occur with traffic management methods. To us, QoE is far more than just traffic shaping.

Dan explained how we collect QoE metrics directly from subscriber traffic, then tie that back in with information we collect directly from the network elements via our poller component. This gives us an enormous data pool that we use to understand the real-world performance of networks. For example, we use this to build performance models so that we can understand what a PMP450m is really capable of (not what the vendor marketing spec sheets say).

We then compare that with the behavior of local APs so we can tell regional operators where there are problems, such as which of their access points have the worst RF conditions. In short, we use data to enable our clients to be proactive, as opposed to just crushing traffic to try and make the network less busy.

Dan also spoke in general about how Preseem can help the various teams at your organization, such as Access Network Operations, Core Network Operations, Support, Sales and Marketing, and Executives and Finance, as well as your subscribers.

Why We Created Automatic Access Point Capacity Management

The problem we’re trying to solve with Automatic Access Point Capacity Management is when overloaded APs deliver a poor subscriber experience. As link utilization goes up, you often see a situation where latency suddenly spikes and “good put” (i.e. how much useful throughput is going through the AP) quickly goes down.

What’s happening there is that packet queues are filling up and are not being managed properly, causing latency to increase. Eventually, packets will also start being dropped, and this leads to a poor experience for your subscribers.

Diagram showing issues with packet queues

We’ve had manual AP shapers in Preseem for quite a while, but the problem with those is right there in the name—manual. It’s time-consuming, especially on large networks, and we found that many Preseem customers only used them to temporarily mitigate the worst APs until they could be fixed.

Why We Don’t Do Artificial Demand Reduction

We’re philosophically opposed to application-specific shaping. Yes, if you artificially reduce demand on an AP, then the experience will temporarily improve on a busy network. However, it doesn’t fix throughput per application, which is something customers really care about and will notice.

It also doesn’t address the real problem, which is AP congestion. The problem isn’t individual applications, the problem is that APs have fundamental limits. As well, application-specific shaping brings up the question of how you go about choosing which applications to shape.

For example, what you decide is “not as important” may be of critical importance to some of your customers. Application shaping is counter to what your subscribers really want. Also, it actually makes your service less competitive against other access technologies like fiber and cable—they abandoned this kind of thing long ago.

Fixed wireless providers need to manage the real fundamental limit in the network—the access points—and not try and make decisions with individual applications all the time. Application-specific shaping is not a long-term solution and puts ISPs in tension with customers (by de-prioritizing gaming updates, for example, you’re going to have some upset gamers on your hands). Typically, traffic will grow 17-20% per year, and if you’re not planning for that, your customers will go elsewhere.

Introducing the Easy Button for AP Capacity Management

To automate access point capacity management, we take a number of factors into account:

  • Understanding the network topology—what AP is the subscriber attached to and what are the services downstream from that CPE?
  • AP performance and RF conditions—e.g. what are the modulation rates on the AP?
  • Real-world performance information—the billions of metrics collected daily by Preseem from hundreds of regional operators worldwide give us the unique ability to determine (for example) the optimal rate that should be sent through a particular AP, and that can then be automated
Automatic AP Capacity Management is the solution to access point congestion

All of the above factors are then rolled into an algorithm that creates an automatic repeated adjustment of the traffic rate on an AP. It will adapt to changes in conditions, and you’ll consistently have great QoE even on busy APs. This will keep your customers happy and also dramatically decrease the amount of labor you’re putting into the process, which means lower operational costs.

One of the major benefits of Automatic Access Point Capacity Management is that it helps improve latency. With this new feature, we’re already seeing up to 25 ms of improvements in latency when a network is idle, loaded, or stressed. Most significantly, the biggest improvements in latency are on stressed networks (e.g. 75%+ of airtime is being used). This is in line with our goal to optimize access points for peak busy hours.

Side note: If you need a refresher on the world of airtime, modulations, link rates, etc. download our free RF eBook to get back up to speed.

Jeremy also walked through real-world examples from customers who’ve been trying out the new feature since January. These slides clearly show how it’s been reducing latency on real networks without reducing throughput. Dan and Jeremy also showed screenshots of how to configure and enable the feature in Penelope, and how to read the graphs it generates.

Our team also discussed what happens when you shape flexible framing, how auto shaping can help expose RF performance problems, and answered questions from the webinar attendees. You can watch the full video below to hear the Q&A portion of the event.

By the way, if you’re not yet a Preseem user and would like to see what Automatic Access Point Capacity Management can do for your network, book a demo with us and we’ll be happy to show you the ropes.

Subscribe to the Preseem Blog Newsletter

Stay in-the-know and get fresh content delivered to your inbox once a month.