Software is only as good as the data it processes. Likewise, log processing is only as good as the logs generated. Perhaps we should look at it this way: good information enables good insights and effective decisions. To gain benefits from capturing, manipulating, and routing log events, we need log events that are clear and precise in meaning. The more directly software can get log events to Fluentd, the lower the overheads and the less opportunity for ambiguity to creep into monitoring.
In the remaining chapters, we will focus on ensuring the log content has maximum value and meaning. In short, what makes a good log event? When are logs oversharing or not sharing enough? Is the context of the log event clear?
Many applications these days use logging frameworks either as part of the language or from third parties. What should we look to gain from such frameworks? What should we look for when choosing a framework? What options exist if there isn’t a framework or a specific means to connect to Fluentd? Fluentd provides a range of additional utilities and libraries that can help, so what is available? These are all questions we’ll engage with in the last part of the book.