We Gave Memory-Optimized Tables a Hash Lookup — Then Tried Pattern Matching Instead

Question

We Gave Memory-Optimized Tables a Hash Lookup — Then Tried Pattern Matching Instead

Chandan Shukla

Say Hey Kid

Points: 669
More actions
July 22, 2025 at 5:00 am

#4625479

Comments posted to this topic are about the item We Gave Memory-Optimized Tables a Hash Lookup — Then Tried Pattern Matching Instead

Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply

Jakub.Janda Default port Points: 1428 More actions · Answer 1

I would like to see the numbers for disk table and pattern matching, should the database use optimistic locking, i.e. READ_COMMITTED_SNAPSHOT. I'd guess the advantage of memory tables should largely disappear.

Chandan Shukla Say Hey Kid Points: 669 More actions · Answer 2

@Jakub Janda.. Adding RCSI still cause below issues:

It doesn’t eliminate latching or I/O stalls.

It adds TempDB pressure (since row versions are stored there).

And it still doesn’t match the raw concurrency and latch-free architecture of memory-optimized tables.