App Performance Management
for the International Official Website of Haodf
Established in 2006, Haodf is a leading Internet healthcare platform. With 11 years of honest operation, Haodf has achieved a leading position in many fields such as graphic and text consultation, telephone consultation, remote outpatient consultation, appointment referral, post-diagnosis management, and family doctors. As of October 2017, Haodf recruited 529,000 doctors from 8241 regular hospitals across China.
Customer pain points
Haodf app lacks effective monitoring means to perceive the system operating status. As a result, operation personnel cannot understand the health status of the app in a timely manner and quickly identify and resolve problems. The response time of the app is excessively long, with symptoms such as longtime page loading or blank screen. The operation department often receives complaints from clients. However, the operation department is often the last to be aware of performance bottlenecks and faults of the app, and has to respond to customer complaints passively, resulting in unresponsive problem resolution. The details are as follows:
Handling of complaints from key users: Haodf attaches great importance to the user experience of doctor users. The churn of doctor-end users is a great loss for Haodf. Each user's complaints and feedback needs to be properly analyzed and resolved.
Network performance analysis: In the app, static resources (such as pictures) are obtained from the CDN, but dynamic resources are retrieved from their source websites and located under multiple domain names. The network performance varies with interfaces. For any domain name or interface involving network performance deterioration, problems must be quickly located and resolved.
App crash analysis: This is a common problem. Each app version update will resolve some known crash problems but introduce new crash problems. The crash rate must be under monitoring and make sure it is converging.
The Cloudwise Synthetic Monitoring monitors the website, webpage performance, and network quality. It covers over 300 IDC nodes worldwide and actively detects the health status of the website, app, and network. It complies with HTTP, HPPTS, Ping, DNS, SMTP, TCP, Traceroute, FTP, UDP, and MTR.
Real user IDs and nicknames are bound through the Cloudwise APM API. When a user makes a complaint, the Cloudwise APM directly searches for the user information through the user ID, enabling active and quick problem handling.
Network performance is analyzed from multiple dimensions including domain name, interface, and region. In addition to the TTFB breakdown (DNS, first byte, ...) inherent in the Cloudwise APM, the performance status of all network requests initiated by this user in the same time period can be retrieved by the user ID. root cause of network problems are located through cross-domain name comparison.
Crash information is fully collected, including mobile phone information, crash track, and crash stack, to help R&D personnel quickly identify the causes of the crash.
The mobile app performance scoring system of the Cloudwise APM scores the platform/device apps by overall performance from the four dimensions: response time, crash rate, HTTP error rate, and network failure rate, to quickly identify vulnerabilities in app user experience and continuously improve the user experience of Haodf mobile app.
User experience data is bound to real user IDs, to analyze app performance from the user perspective, accurately recover the user experience of individual users and the details, and efficiently handle user feedbacks.
IT personnel can quickly determine the impact scope of problems, and quickly diagnose, re-emergence, and resolve the problems, improving operation efficiency significantly.