Also from Fair here. A lot of the cost savings were because we could decrease the size of our elasticsearch cluster. This was mostly due to the flattening of t

Also from Fair here. A lot of the cost savings were because we could decrease the size of our elasticsearch cluster. This was mostly due to the flattening of the elasticsearch document schema, which improved performance dramatically but also allowed us to run significantly fewer ES nodes.