Clustered indexes are a feature of InMemory, that help you speed up certain queries, when you filter by a common dimension.
Suppose you have a main fact table called SalesTable, and a date dimension called InvoiceDate, you could create a clustered index, by running the following statement in IRDBImport or IRDB-SQL
CREATE CLUSTERED INDEX ON SalesTable(InvoiceDate)
In our earlier builds, our tool would search out range filters or simple list of equality expressions involving the clustered index, on the first table ( or driving table). This would then speed up queries by only scanning those records.
We have enhanced our clustered index support. We can now analyze joins (LEFT or INNER ) involving the clustered column. We will analyze the where clause, and try and extract a filter that just applies to the joined table. We will then execute a simplified query on the joined table, and then use it to generate a subset of the records on the driving table, to query.
This means if you have a lookup table that joins on the clustered column , and you filter on attributes in that table, it should still be able to use the clustered index to speed your query.