In dramatic fashion, I’m revisiting something from this post about stats updates.
It’s a quick post, because uh… Well. Pick a reason.
Get In Gear
Follow along as I repeat all the steps in the linked post to:
- Load > 2 billion rows into a table
- Create a stats object on every column
- Load enough new data to trigger a stats refresh
- Query the table to trigger the stats refresh
Except this time, I’m adding a mAxDoP 1 hint to it:
WHERE UserId = 138
AND PostId = 138
AND BountyAmount = 138
AND VoteTypeId = 138
AND CreationDate = 138
Here’s Where Things Get Interesting
Our MaXdOp 1 query registers nearly the same amount of time on stats updates and parallelism.
But our plan is indeed serial. Because we told it to be.
By setting maxDOP to 1.
So, if you’re out there in the world wondering why this crazy kinda thing goes down, here’s one explanation.
Are there others? Probably.
But you’ll have to find out by setting MAXdop to 1 on your own.
Thanks for reading!