Do You Have a HEAP of Hurting?

Episode Quotes

“A heap is a table that doesn’t have a clustered index. What that really means is a table that doesn’t have a primary key. There are times when the primary key and clustered index could be different, but more often than not, they’re the same.”

“The case I try to make when I talk about them is that there might not be as many use cases for heaps as we assume. They’re great for write-only workloads doesn’t mean that they’re great for workloads where you mostly write and sometimes do other things.”

“If it’s a situation where the table is going to just kind of be loaded and then just radically, completely and totally transformed and destroyed and written somewhere else, that could possibly be a use case for a heap.”

Listen to Learn

00:40     Intro
01:09     Compañero Shout-Outs
01:45     Conference
02:03     Intro to the guest and topic
05:09     Heaps are great if your workload is write-only
06:15     Deadlocks happen & Definition of “small table”
07:35     Heaps with ETL vs ELT
11:53     The SELECT INTO use case
12:58     Are we just skirting around heaps when we’re using an identity column?
14:19     Scenarios where you might use a heap
15:48     GUID as your primary key?
16:36     What about hot page scenarios?
19:25     It’s an awful idea to create a clustered index and then drop it
22:18     A heap may be better than a really poorly designed clustered index
24:32     It is okay to just stay away from heaps, especially if you’re new to execution plans
28:11     SQL Family Questions
31:22     Closing Thoughts