I was out for hiking last weekend. So I was not able to write #41.
http://www.confluent.io/blog/apache-kafka-samza-and-the-unix-philosophy-of-distributed-data
Martin Kleppmenn talks about the philosophy behind using Kafka in distributed system. Comparing to the Unix, Kafa is the new pipe.
His blog has lots of good read.
http://sigops.org/sosp/sosp15/current/2015-Monterey/008-tang-online.pdf
Facebooks’ configuration management tools. They write code as configs and store the configs generated by code in Git. Interestingly, this is how they solve the problem of configuration dependencies: 1. The code to generate configs has dependencies. 2. An interpreter will grab the dependency from the code and figure out the dependency.
Lots of stuff in this paper.