QReport vs Beacon: A Frontend‑Centric Real‑Time Data Reporting Solution

In 2014 the author joined Qunar as a front‑end engineer and, while using the Beacon data analysis system, encountered many shortcomings that motivated the design and development of a new system called QReport.

Beacon is the most widely used data analysis platform at Qunar; it records all event logs as files and provides analysis mainly through tabular reports.

QReport differs by offering a real‑time log stream in its logging service, enabling immediate analysis and a visualized dashboard that presents data graphically as well as raw logs.

Unlike Beacon, which uses a page‑specific sid that must be manually maintained in front‑end code, QReport uses the page URL as the unique identifier, allowing the tracking script to obtain it automatically without extra mapping work.

For cases where URL differentiation is insufficient, QReport also supports a page‑id mapping mechanism.

QReport simplifies instrumentation: a single script tag is added to pages, and updates are delivered via negotiated caching, eliminating the need for developers to modify code for each new page.

For page‑flow analysis, Beacon relies on the HTTP refer header, which often loses data (e.g., https→http) and cannot distinguish multiple tabs. QReport generates a random pid for each page and propagates it to the next page, enabling precise upstream‑downstream tracking even across tabs.

Beacon processes logs offline, so data becomes visible only the next day, which is unsuitable for time‑sensitive operations. QReport uses Kafka to provide a real‑time log stream, stores analysis results in MySQL, and displays them instantly on its front‑end platform.

Beacon’s metrics are limited to page views (PV). QReport expands the metric set to include PV, AC (user actions), PF (page performance), ST (stay time), ER (script errors), and EX (custom events), and provides unified analysis for all of them.

Simpler instrumentation: a single script tag handles tracking.

Real‑time visibility of data after deployment.

Complete page‑flow tracking with a new pid‑based method.

Richer statistical indicators beyond basic PV.

The QReport system now supports real‑time PV and UV, page flow, upstream/downstream relationships, error monitoring, click heatmaps, and more.

PV, UV metrics

Click heatmap

Error statistics

Finally, the author invites readers to try QReport and provides contact information (Qtalk ID: hongwei.xu).