How TrafficPeak Helped Provide a Superior Football Viewing Experience
On a big game day, troubleshooting issues and potential incidents in real time is critical for making sure the event is successful. TrafficPeak gives us real-time visibility into our media customers’ data, which we use to balance traffic across geographies and surpass viewers’ expectations of a smooth, reliable experience.
Business challenge
At Akamai, large sporting events, especially the biggest football game of the year, are some of the most important events of the year. The largest media companies use our solutions to stream the game to viewers worldwide. However, to ensure each viewer enjoys the game without struggling with buffering and outages, we need to collect detailed data in real time.
This year’s final football game was bigger than ever. With nearly 124 million average viewers, it was reportedly the most-watched live broadcast event since the Apollo 11 moon landing. Streaming media and broadcasters are among Akamai’s biggest customers, which meant making sure they shined during the event was our top priority.
The need to quickly mitigate performance issues
For one media provider, our team at Akamai needed to query CDN logs, which required retrieving data in seconds. Unfortunately for us, collecting and analyzing DataStream 2 and CMCD data was taking 20 minutes under peak traffic volumes. For the big game and any other live event, 20 minutes is too long.
To mitigate performance and security issues quickly, before viewers felt the impact, we needed data insights in real time. We needed to collect, analyze, and visualize data, such as geography, city, state, autonomous system number (ASN), and server details, in real time and in one view to more easily find and fix errors.
If we could instantly see which issues were unique to one provider and which ones were spread across ASN networks, we could quickly make mitigation decisions, such as moving traffic to different regions to bypass impacted transits.
For postgame analysis, we also wanted to drill into the performance outliers and see their sources. That way, we could get a clear picture of their impact and make data-driven mitigation decisions in the future.
The big game
Since TrafficPeak is our Akamai-managed observability service powered by Hydrolix, it was a no-brainer to use it for the big game. The service is easy to deploy, even with customization.
We built out visualization around the top ASNs, response codes, and errors. We viewed delivery performance metrics and could see degradation in certain regions. For example, it was easy to filter by city to see if the performance looked on point or if outliers existed and then apply fixes before viewers were impacted.
For us, TrafficPeak provided the most value in three areas: ease of customization, real-time data, and architecture.
Ease of customization
As a technical person, writing complex queries does come easily to me. Even so, with TrafficPeak, the ability to customize any data insights into custom dashboards is an amazing and impressive capability.
Real-time data
Although similar solutions exist, like ELK Stack, for ingesting, processing, and visualizing data, they are not as quick as TrafficPeak. At peak traffic time, we were collecting almost 11 million records per second, and slicing and querying the data took subseconds. It is difficult to replicate that speed with other tools.
Architecture
TrafficPeak is designed to be inside Kubernetes. As a Kubernetes user, I understand how to scale and distribute the service. Its design is ideal for cloud deployments. It’s clear that a lot of thought went into the software design and back-end architecture.
The results
The big game was a successful event. We collected 53 billion records and transformed 41 terabytes of raw data into 5.76 terabytes of compressed data stored. We had low error counts, which I attribute to TrafficPeak providing real-time visibility into errors as they occurred so that we could improve them instantly.
In addition, getting visibility into all our data was really helpful. In general, with live events, it’s hard to see and understand what’s happening in real time. Our media customers need metrics such as the average bitrate so they can ensure that most users are on the top end of the rendition ladder. TrafficPeak gave us those metrics so we could mitigate them in real time.
After the event, we used TrafficPeak to dig into any performance issues that occurred. The findings showed areas where we can make improvements for next time, such as rewriting filters to better spot similar issues. Now we know how to build the right filters for future events so that those same issues don’t happen again.
Reap the benefits
Use TrafficPeak to uncover and address performance and security issues preemptively. Safeguard your brand’s trust and desirability while increasing your bottom line.
