In today’s competitive market, customer loyalty is a priceless asset for Internet Service Providers (ISPs). Your loyal customers not only bring a steady revenue stream but also act as brand ambassadors, driving new business for you through word-of-mouth and social sharing. This fact is illustrated in the McKinsey & Company study results below which indeed shows that the user’s experience of the network is the major source of churn and drives customer net promoter scores (NPS).
Figure 1 User Experience is the major source of churn and drives customer NPS
As we see from the study, the network and the user’s experience of the network are the top experience factors for roughly 40% of the subscribers. Similarly, Amazon has found that every 100ms of added page load time results in a 1% decrease in sales. Optimizing the user experience for your customers is crucial to remain competitive. We will focus in this blog on optimizing the user experience for web browsing, but the techniques are similar for other types of customer traffic over the Internet.
Obtaining Visibility
The first step to optimize the Internet User Experience is to gain visibility of the current status for your customers. In an earlier blog on user experience monitoring , we noted that you need to monitor the web browsing user experience by measuring both domain name service and web page download speeds. A suitable strategy would be to identify the top 30 or top 50 websites that your customers visit and to run synthetic tests emulating web browsing towards these sites.
Once you have identified potential problematic sites, you can optimize your own network with a carefully designed and implemented TWAMP-based active monitoring strategy. However, when it comes to web browsing, the issues many times are outside of your direct control. The performance issues may not be in your own network, but rather it may be the Internet that is performing badly. To optimize the performance over the Internet, we need to look at how you can optimize the traffic over your transit networks to get the best possible performance for your customers.
Understanding Transit Networks
Transit networks are the backbone of the Internet. They connect the different ISPs together by providing a platform to exchange data packets between networks. An ISP typically selects multiple transit network providers that agree to take traffic originating from the ISP and route it towards a destination over other transit networks until it reaches the destination. Each of these transit networks forms an independent domain called AS (Autonomous System). The Border Gateway Protocol (BGP) is used to distribute the different possible routes through the different AS domains and by modifying the policies used in BGP you can affect where your traffic is routed when it exits your domain.
Selecting the best Transit Network
To select the best transit network, the operator can utilize active monitoring towards the selected Top-30 or Top-50 websites and monitor the end-to-end performance across all potential exit points from the network. It is then possible to direct traffic towards a particular site by changing the BGP policies to prefer that path over the other available paths, see the picture below:
Assume Path 1 has more packet loss and/or longer delays for the packets. We will then change our BGP policies to prefer Path 2, even though the packets travel over more AS domains than for Path 1.
This strategy is simple and straightforward. For each of your top sites, one of the transit networks will perform best. You do not really need to know the details of each of the paths, it is sufficient that you monitor the quality of each of the available paths and select the best path for the target. Please note that it does not always need to be the same transit network that performs best for all targets. This is totally dependent on what sort of route the packet must take through the Internet to reach the target. Therefore, it is important that you monitor all the sites of interest through all potential AS domains.
Selecting the best Path
But how do you select the best path toward a particular destination? As we all know, the conditions on the Internet fluctuate like the weather, so a particular measurement at a particular time may not represent the overall quality of a particular path. To find the best path, you will need a monitoring system with analytics capabilities that can summarize the quality over a longer time (hour, day, week) where you can define thresholds that define acceptable quality KPIs. The analytics system will then summarize the percentage of time that the path met the set quality criteria over the entire measurement period and how big percentage of the time it did not. By selecting the transit network that offers you the overall best quality, you can optimize the user experience long-term for your customers.
Proactively Resolving Issues
Finally, once you have your measurements and analytics in place, you need to turn from a reactive mode to a proactive mode. Rather than to wait for your customers to come back to you to complain about bad Internet performance, you can actively monitor the performance of your top sites and optimize your BGP settings accordingly before the quality drops to critical levels. By utilizing warning levels and automatic notifications, your active monitoring solution can inform you of sites that are about to become problematic and you have time to fix the issues before your customers react.
Conclusion
Optimizing Internet performance for your customers across multiple transit networks is a challenge that requires a combination of strategic planning, technical expertise, and an advanced active monitoring solution covering both TWAMP and user experience tools to get the needed visibility into your network and the Internet. By implementing the right mix of technologies and best practices, you can significantly enhance the speed, reliability, and overall performance of your offered services and keep your customers happy and loyal.