Slides about FoundationDB

(Amirouche) #1

I published some slides about FoundationDB based on a presentation I have done in my company.

I had the following questions:

  • How does it compare to python OrderedDict (which is the behavior of python’s dict in Python 3.7). FDB key-value space is lexicographically ordered based on the comparison of the keys. It’s similar to this SortedDict. It store bytes but provides a packing function for python data types that preserve ordering.

  • How does it compare to mongodb 3.2 / 3.4 and the upcoming 4.0 / 4.2. I think FDB is easier to operate and offers more / better guarantees but that’s a lot of hope more than actual experience.

  • ElasticSearch: in our case ES is used as an exact match and date range lookups and we use map/reduce for aggrgation. The advantage of FDB is that it doesn’t require to spread documents from the same collection into different indices / buckets (otherwise ES index becomes too big). Is that correct?

  • REDIS, that is a very different database and I replied that REDIS has still a place in our infrastructure (even if I won’t use REDIS in my personal projects)

Also I created a tiny layer called OMGNO using official Python bindings that one can use to do experiments.

edit: fixed omgno link