A typical collection is with less than 5 fields. There is one Genny benchmark with 40 fields (VariadicAggregateExpressions) which doesn’t run in Bonsai M2. 

In mongo-perf a number of tests use the ‘largeDoc’ with 23 fields, including nested arrays and documents. MatchManyFieldsFromBigDocument has 50 fields (the same strings) and 1000 documents.

We will need to add a benchmark for collections with at least 100 fields of mixed types.