Understanding the quality of web browsing enjoyed by users is
key to optimize services and keep users’ loyalty. This is crucial for
Internet Service Providers (ISPs) to anticipate problems. Quality
is subjective, and the complexity of today’s pages challenges its
measurement. OnLoad time and SpeedIndex are notable attempts
to quantify web performance. However, these metrics are computed
using browser instrumentation and, thus, are not available to ISPs.
PAIN (PAssive INdicator) is an automatic system to observe the
performance of web pages at ISPs. It leverages passive flow-level
and DNS measurements which are still available in the network
despite the deployment of HTTPS. With unsupervised learning,
PAIN automatically creates a model from the timeline of requests
issued by browsers to render web pages, and uses it to analyze
the web performance in real-time. We compare PAIN to indicators
based on in-browser instrumentation and find strong correlations
between the approaches. It reflects worsening network conditions
and provides visibility into web performance for ISPs.
