This paper mainly introduces the technical practice of using volcano engine APMPlus to greatly improve the stability of application performance. On The night of February 24, MARS TALK invited the engineers from Volcano Engine APMPlus and Mei Fan to share their technical solutions and implementation details. Sign up for the event group now and you’ll have a chance to get the VR all-in-one Pico Neo3!

I. Business scenarios of “Beauty”

Meizhuang is the main product of Nanjing Blue Whale Network Technology Co., LTD. It focuses on graphic creation and sharing in the circle of friends, and is positioned as an Internet content community for “post-confused” to express themselves, make friends with friends, and learn and improve.

Business scale of “Beauty Chapter” :

Since its official launch in July 2015, the APP has accumulated 200 million registered users, 40 million creative users, 12 million monthly active users and 180,000 new works per day. Since its establishment, it has obtained multiple rounds of financing of 180 million yuan from Tencent Shuangbai Plan, Jingwei Venture Capital, Zhenfund, Mango Cultural Innovation Fund and other first-line investment institutions, with an estimated value of 1 billion yuan. In October 2021, APP 7.0 will be released.

Ii. Business pain points and problems of “The United States”

  • “The United States” business status quo formation of pain points
  • Rapid growth of users: the scale of users grows rapidly, and the number of creative users exceeds 40 million
  • Massive content deposition: As a strong interest content community, Meipuan has a massive UGC graphic and video content deposition
  • Multi-product functions: hot list of topics, online classes, pen Pal letters and other contents

Because of these reasons, the amount of information carried by the application platform increases greatly, which makes the application more difficult to run and threatens its stability.

  • Problems and challenges encountered in “Beauty”

For mobile App developers, the most basic and most concerned issues are the stability of the application and user experience. Crash and lag are the key factors affecting application stability and user experience. Including Java crash, Native crash, ANR, stuck and other problem types. With the rich functions and the rapid growth of users, the optimization and treatment of OOM and lag are becoming more and more important.

3. Mars-apmplus solutions

Product advantage

APMPlus is the performance monitoring product of MARS, a volcano engine application development suite. With advanced data acquisition and monitoring technology, APMPlus provides full-link application performance monitoring services for enterprises, helping enterprises improve the efficiency of troubleshooting and solving abnormal problems

Platform based on the analysis of huge amounts of data aggregation, and can help customers find more kind of unusual problems, and promptly report to the police, do distribution processing, platform provides rich attribution ability at the same time, including but not limited to, abnormal analysis, multidimensional analysis, custom reports, single-point log query and so on, combined with flexible capacity report can understand the trend of all kinds of indicators. APM Plus application performance monitoring has served a number of mobile apps with large scale users, such as Douyin and Toutiao.

A brief introduction to the “Beautiful Chapter” solution

In view of the stability problem of The Us App, we provide different types of exception monitoring, such as Java crash, Native crash and ANR, with high capture success rate and extremely rich attribution. For performance, we provide startup monitoring, page monitoring, lag monitoring and other fluency monitoring, as well as memory, CPU, power and other resource consumption monitoring. It also provides network time and exception monitoring, as well as powerful single point analysis and log recovery capabilities to analyze single point problems. Has the event custom report for individual monitoring, powerful kanban for multi-dimensional display. Data collection Provides sampling and switch configuration to meet users’ demands on data volume and cost, and provides basic alarm capability to report abnormal problems in real time. Provides WebView pages and Flutter monitoring for cross-platform solutions. Rich capabilities to meet customers’ demands for comprehensive App performance monitoring.

Plan highlights

  • The Java OOM monitor automatically analyzes the entire process to locate Java memory problems. Leakage chain, leakage size at a glance
  • ANR uses signal-based capture scheme, which saves system resources and has high accuracy. It is the only way to provide the field message scheduling diagram and highly restore the field main thread blocking situation
  • The on-site restore capability to truly solve Native(C/C++) crashes provides the most valuable Tombstone for fine restore sites. Complete display of crash thread process information, signal information, register information, restore crash scene assembly instructions, detailed MAPS, FD and memory information
  • The elog provides a high-performance log library, which ensures high data stability and performance and ensures high recovery of on-site service information

Iv. Implementation effect

Using the MarS-APMPlus application performance monitoring product and implementing the above solution, the results were significant and the performance stability of the application was significantly improved

Caton drop:

OOM down:

  • The Android Java OOM dropped more than 80%, from 1.5% in August to 0.1% at the end of December
  • The Android user lag rate dropped by more than 80%, from 11% in August to 2% at the end of December
  • The iOS OOM dropped more than 80%, from 1.3% in August to 0.2% in December
  • The iOS lag rate decreased by more than 90%, from 1.1% in August to 0.1% in December

Check out MARS TALK 04 for more practical details! At that time, we invited The volcano Engine APMPlus R&D engineer Wang Tao and The United States research and development engineer Ye Wanwan lecturer, Online share “APMPlus Java OOM Attribution scheme based on Hprof file” and “Optimization practice based on Mars-APMPlus Performance monitoring tool” and other technical dry goods.

⏰ Live: 20:00-21:30, Thursday, February 24

💡 Event format: live online

🙋 : Scan the qr code below to enter the group

As the first MARS TALK of the New Year, we have prepared the latest VR all-in-one machine — Pico Neo3 as a gift for you, in addition to logitech M720 Bluetooth mouse, fascia gun and byte peripheral gifts for you to pick up. Pass by don’t miss yo!

👉 Click here to learn about APMPlus