How to Trace the Root Cause of Application Performance Issues Across the Full Stack

During DevOps we often encounter these problems: How to quickly pinpoint the cause when an app becomes slow. How to distinguish whether website slowness is due to network issues or backend code. How to identify if severe page performance loss stems from page structure or resource request delays. How to accurately trace the root cause of complex cross‑system performance issues.

On September 23‑24, 2016, the GOPS2016 Global Operations Conference was held in Shanghai. Vice President of R&D at Tingyun, Liao Xiongjie, delivered a talk titled “Strategic Planning – Tracing the Root Cause of Application Performance Issues,” discussing performance problems in complex delivery chains and how to trace them from app, browser to server, especially under micro‑service architectures. A post‑event interview followed.

▲ Liao Xiongjie, Vice President of R&D, Tingyun

In the Internet era, enterprises rely heavily on IT systems, making user‑experience‑centric operations essential. Personalized user demands push companies to seek long‑term success of applications and continuous delivery capabilities, prompting a focus on user experience.

Meanwhile, the proliferation of mobile Internet and cloud computing has reshaped business models; the "cloud‑manage‑endpoint" architecture is becoming the new norm, and mobile apps dominate the battlefield. Large internet giants dominate the market, leaving many companies struggling to maintain application performance despite heavy investment.

Application Performance Management (APM) is defined by Baidu Baike as a newer network‑management discipline that monitors and optimizes critical business applications, improving reliability, quality, user service, and reducing total IT cost.

APM Is the Industry Standard

Liao noted that Tingyun celebrated its tenth anniversary, during which its APM journey evolved from client‑side to server‑side, from external to internal, from PC to mobile, and from proactive to passive monitoring. Early products focused on proactive web‑app testing (Tingyun Network) addressing web‑app experience, IDC/CDN/cloud selection, network optimization, and competitor performance comparison.

Since 2009, Tingyun explored mobile Internet APM, launching simulated mobile monitoring, real‑device monitoring, and the Tingyun App product to address performance issues in weak mobile networks, diverse devices, and operating systems.

Client‑side APM revealed network and terminal‑layer issues, but problems behind firewalls remained unsolvable, prompting server‑side APM development in 2013 and the release of Tingyun Server in 2014, which monitors code, databases, NoSQL, and other backend services. In 2015, Tingyun Browser was introduced for browser‑side monitoring.

Full‑Stack Tracing – Connecting the Entire Application Flow

Liao questioned whether a single APM model can solve all monitoring challenges in today’s increasingly complex operations.

Modern development spans servers, mobile devices, browsers, and even mini‑programs within platforms like WeChat, creating intricate interaction chains.

How can we monitor quickly and, when issues arise, precisely locate whether they stem from network, mobile, browser, or server performance? Tingyun’s full‑stack tracing was born from this need.

Full‑stack tracing aggregates fragmented problems and serially pinpoints their origins, enabling both developers and operations engineers to locate issues directly and resolve them efficiently.

Today, Tingyun’s solution covers client‑to‑server, browser‑to‑server, network‑to‑server, and inter‑server tracing, helping DevOps rapidly troubleshoot performance across diverse business logic.

Tingyun’s Focus and the Future of APM

Many industries are moving toward SaaS, and APM is no exception. Migrating to the cloud and adopting elastic deployments are inevitable under the trend of internet‑centric applications. APM providers must continuously refine technology and services to adapt to cloud environments.

Big data advances will also drive APM evolution. As SaaS APM services collect massive real‑user performance data, leveraging machine learning and data mining to extract value—such as predicting potential performance failures and offering proactive assessments—becomes crucial.

In the coming years, Tingyun will continue focusing on APM R&D, emphasizing mobile, cloud, and big data. Mobile support will expand to more device types, platforms, H5, and hybrid apps.

On the cloud side, Tingyun will collaborate closely with major cloud providers to deliver simple, user‑friendly cloud‑based APM products across multiple platforms.

In data analysis, Tingyun plans deeper mining of collected massive datasets, delivering richer industry performance reports and data services, exemplified by products like Tingyun Controller and Tingyun API Index.