You have now built your application and instituted various processes and mechanisms in anticipation of changes over its lifetime. An additional, but no less essential, process that you must have in place and carry out from time to time is monitoring and performance measurement. Any production application you build for real-world users and institutions must meet certain performance goals to be useful to its users, and by implication, the application's stakeholders. Therefore, understanding how your application performs and trying to improve its performance is a key maintenance task; any dereliction in this task may result in your application having a short shelf life.
The art (and science) of system performance measurement and analytics is a broad and extensive set of topics, and it is not our intention to cover these topics deeply or exhaustively in this book. To obtain such a coverage, the interested reader is encouraged to read other canonical texts on the topic (for example, https://www.amazon.com/Systems-Performance-Enterprise-Brendan-Gregg/dp/0133390098.) Instead, we will offer a preview of what performance measurement and gaining insight into a blockchain application entails, and offer some hints and suggestions about the tools and techniques a developer or system administrator can utilize for these purposes.
Broadly speaking, systems maintenance for performance involves three, roughly sequential, categories of tasks, though these tasks can collectively repeat in cycles over the lifetime of a system:
- Observation and measurement
- Evaluation (or analysis) and gaining insight (or understanding)
- Restructuring, redesign, or reimplementation for improvement
In our discussion in this section, we will mainly focus on some aspects of the following:
- What is important to measure in a Fabric application
- The mechanisms a Fabric application developer or administrator can use for measurement
- The performance-inhibiting aspects of Fabric that an application designers and developers should be aware of