Following on from Part 1 – An Introduction to In-Memory OLTP this post will focus on the new types of indexes that are supported. I mentioned in part 1 that clustered indexes are not supported for memory optimised objects and that there can only be one unique index (the primary key). In addition the following should also be taken into consideration:- A maximum of 8 indexes on one table Indexes are never written to disk, they are entirely in memory Created when the table is created (in-line declaration) In-Memory OLTP introduces two new types of indexes, Hash and Range. These are very different from the existing disk based indexes we know well so I’ll go through the internal structures of each type and when to use them. Hash Indexes Hash indexes utilise a hash function to each value in the index row in order to sort the records in the index into “buckets” Range Indexes Range indexes utilise a internal structure known as a Bw-tree.
↧