Education
An Internet Education Company in China
About The Client
The client is a high-tech company that runs an Internet education product created by an elite team. It is mainly engaged in Internet education for online tutoring in primary and secondary schools. It focuses on providing multidisciplinary online learning services for primary and secondary school students aged 9 to 15. It uses the Internet and the “assessment and training” system independently researched and developed by big data and new technology to accurately measure the learning level of students, and intelligently push the most appropriate exercises according to the learning level, so that parents, students and teachers can understand the learning status of students in real time. , develop a targeted plan.
Project Features
The system needs to calculate the student ranking in the entire system which makes a great impact on system resources and system response time due to various reasons.
All students’ assignments need to be uploaded to the server and it has a high demand for system stability and responsiveness.
The system not only needs to be compatible with users from different provinces and different bandwidths, but also to deal with thousands of pictures and test resources at same time, and it has a high demand for network bandwidth transmission capability.
The system is deployed on the public cloud platform, so also purchased the service components of the public cloud platform. However, the inner workings of such components is not completely known by the outside world, so it is not known whether it will become a performance bottleneck.
Service Value
Used automated performance testing tools to simulate request from the Web end.
Identified the potential bottlenecks which may lead to CPU deadlock, reduced potential risks successfully by adjusting the execution time of backend tasks.
Identified the problem of insufficient disk capacity due to the improper server logging.
Founded seven slow SQL statements in the database, and largely improved the response time after optimizing.
The number of concurrency under the main operating scenario increased to 8000, comparing with initially 500 before the tuning .
The number of concurrent visits under the major scenarios increased to 1100, comparing with initially 500 before the tuning.