You’re struggling to get your application to perform optimally. What tools can help you get there?

I often use: Chrome DevTools including the Performance and Memory tabs for analyzing CPU and memory usage, and the Lighthouse tool for auditing web page performance. And if you working with React, you can consider about React DevTools, it provides insights into component rendering, state changes, and performance optimizations.

Two main types of code analysis tools: 1. Static code analysis 2. Dynamic code analysis Popular analysis tools: - SonarQube: An open-source platform that provides static code analysis, code reviews, and project management features. - Fortify: A commercial static code analysis tool that provides a wide range of features, including security vulnerability scanning, code quality analysis, and compliance checking. - PMD: An open-source static code analysis tool that focuses on Java code. - Code Climate: A commercial code analysis tool that provides a variety of features, including code quality analysis, code reviews, and project management features. - ReSharper: A commercial code analysis tool that is integrated into the Visual Studio IDE.

Popular Profiling Tools: - Visual Studio Profiler (Windows): Integrated into Visual Studio, it offers CPU, memory, and allocation profiling for various programming languages. - Xcode Instruments (macOS): Built-in tool for macOS development, it provides CPU, memory, network, and energy profiling functionalities. - Java VisualVM: Open-source tool for profiling Java applications, offering CPU, memory, and thread profiling capabilities. - JProfiler: Commercial profiler offering comprehensive profiling features for Java applications, including CPU, memory, heap, and thread profiling. - Valgrind (Linux): Open-source memory debugger and profiler primarily used for Linux systems, offering memory leak detection and profiling capabilities.

Types of Load Testing Tools: - Open-source tools: Freely available options like Apache JMeter and Gatling offer powerful features for load testing various applications. - Commercial tools: Paid solutions like LoadRunner and LoadView provide additional features like real browser testing, advanced reporting, and cloud-based testing infrastructure.

L'observabilité est un sujet à échanger avec les développeurs dès la conception de l'application. Les KPI doivent être spécifiés, les services associés doivent être identifiés afin de générer les éléments dans la log nécessaires à leur mise en place. Pour que les analyses soient possibles, certains attributs des requêtes et réponses doivent remonter dans la log, dans le respect de la GDPR. Le volume de log doit être considéré afin de garder un bon niveau de performance. Le pipeline d'intégration et surtout la suppression des logs doivent aussi être organisés dès la conception. Le SRE pilote ces sujets avec les équipes Dev & Ops.

Adding to it: 1. Monitor the response times for services and utilize APM tools like Dynatrace or AppDynamics to trace the requests to breakdown costs of execution at different steps of request processing or response building 2. Analyze code for way objects are designed to hold and process data to avoid time lapse in building them 3. Are the resources used for storing/retriving can utilize cloud availability (regions) to avoid latency to access 4. Looking for possibilities to perform caching or async processing that can help avoid wait times

Above shared are some ways to figure out if your application lacks optimal performance. I would suggest some ways you can fix your performance: 1. Maintain connection pooling, it means you should not open a new connection everytime you want to query a database. Some modern frameworks do it for you. but make sure its present ;) 2. Check for bottlenecks in your code i.e is there any code which is causing deadlocks or your mutexes to wait indefinitely. 3. Dont overstuff your apps with new and shiny object technologies just so you can satisfy your inner developer that you have made the application complex enough ??. For example, if you see that there is not huge scaling and descaling in your apps, theres no point using kubernetes in it.

There are many reasons why software does not perform or scale. The first step is instrumentation to understand under what scenarios it isn't performing and where time is spent. Tools like Grafana are good for getting an overall view of your software and are helpful when dealing with performance issues for use cases that span multiple services. Adding instrumentation code may be necessary to break down where time is spent and to know where to focus. Code-level issues should be easier to spot after this. Bottlenecks stemming from the design will require more work and investigation. The lesson to take away from this - add more performance test scenarios.