Logging
When working with highly distributed containers, you may have 100 instances of your application running, rather than one or two. This means that if you need to grep your log files, you will be doing this over hundreds of files instead of just a couple. In addition, Docker-based applications should be stateless and the scheduler may be moving them around on multiple hosts. This adds an extra layer of complexity to manage. To save the trouble, the best way to solves this problem is not to write the logs to disk in the first place. A distributed logging store, such as an ELK stack, or an SaaS platform, such as Logmatic or Loggly, solve this problem for us and gives us a fantastic insight into the health and operating condition of our system. Regarding the cost, you will most likely find that one of the SaaS providers is cheaper than running and maintaining your ELK stack. However, your security needs may not always allow this. Retention is also an interesting problem while looking at...
