While checking once again my sources before publishing a “transcript” of my seminars at FOSDEM and cfgmgmtcamp, I found out that something in the way I presented LinkedIn’s ENC didn’t sound right. I contacted Mike Svoboda and, unfortunately, he confirmed that I actually mixed up two different things.

What was right: their ENC solutions is actually based on Range, as Mike himself explains in this post in the help-cfengine mailing list. It is also true that Range collects information from many different sources, including sources defined outside of the System Operations group, and makes that information available to nodes. In turn, the nodes  use that information for classification by means of bash and python scripts.

What was wrong: the system illustrated by Mike in his seminar “Leveraging In-Memory Key Value Stores for Large-Scale Operations” is separate, and someway complementary, to the ENC system, and is implemented using Redis. Mike has published the details about it in the help-cfengine mailing list.

Apologies for the mistake 😦