In a recent post on StreamingMedia.com, Dan Rayburn shares his insights gained from an interview with David Drai, founder of Anodot. Drai spoke with Rayburn about the challenges that CDN operators face and how machine learning can help. Today’s CDN providers still have not found a way to optimize visibility in order to correct basic issues including increased HTTP errors, IO access or rates of cache churn. According to Rayburn, this can delay upgrade releases and have a direct impact on a CDN provider’s pace of innovation.
One challenge is the quality of log analytics. CDN providers typically use in-house systems to collect billions of transaction logs. Older legacy systems may run queries but they generally cannot keep up with the magnitude of data and are unable to provide results in real time. Outdated data skew results of critical reports.
Drai says that “when dealing with CDN issues, time is of the essence. The user download rate of one of our gaming customers at Contendo decreased by 10-15% due to an issue that took us almost a week to detect. In the world of CDN, that kind of delay can significantly damage the CDN provider’s reputation.”
As data analytics have evolved over the last decade, we’ve seen a shift to flexible, agile big data solutions versus traditional slow and bulky systems. Monitoring solutions will be next, and Rayburn suggests that the next phase for analytics is machine learning.
“The ultimate solution is to automate data-based learning, then develop insights and make relevant predictions.,” Rayburn forecasts. The current reliance on human analysis will be replaced by advanced machine learning solutions that can run pattern recognition algorithms and predictive analytics, like the Anodot team is developing. According to Drai, the key is “zero human configurations.”