Service Level Loggin/Tracing
- by Ahsan Alam
We all love to develop services, right? First timers want to learn technologies like WCF and Web Services. Some simply want to build services; whereas, others may find services as natural architectural decision for particular systems. Whatever the reason might be, services are commonly used in building wide range of systems.
Developers often encapsulates various functionality (small or big) within one or more services, and expose them for multiple applications. Sometimes from day one (and definitely over time) these services may evolve into a set of black boxes. Services or not, black boxes or not, issues and exceptions are sometimes hard to avoid, especially in highly evolving and transactional systems. We can try to be methodical with our unit testing, QA and overall process; but we may not be able to avoid some type of system issues. When issues arise from one or more highly transactional services, it becomes necessary to resolve them very quickly. When systems handle thousands of transaction in matter of hours, some issues may not surface immediately. That is when service level logging becomes very useful. Technologies such as WCF, allow us to enable service level tracing with minimal effort; but that may not provide us with complete picture. Developers may need to add tracing within critical areas of the code with various degrees of verbosity. Programmer can always utilize some logging framework such as the 'Logging Application Block' to get the job done. It may seem overkill sometimes; but I have noticed from my experience that service level logging helps programmer trace many issues very quickly.
