Logging is great. However, in a distributed system, you will want to feed all of your logs and exceptions into a single location, where you can easily analyze them and generate alerts. One potential option for this is Logstash (https://www.elastic.co/products/logstash), which we fleetingly mentioned in Chapter 4, Measuring Performance Bottlenecks.
If you prefer a more modular approach and want to record performance counters and metrics, then there is StatsD, which listens for UDP packets and pushes to Graphite for storage and graphing. You can get it at https://github.com/etsy/statsd and there are a few .NET clients listed on the wiki, along with a sample C# code in the main repository. Another open source option for collecting ...
