Shop Talk: 2021-04-12

The Recording

The Panelists

  • Kevin Feasel
  • Mike Chrestensen

Notes: Questions and Topics

Questions of Critical Importance

Tonight’s episode was a fun topic where I essentially gave a survey but without collecting responses. Because that’s how I roll. Here were the questions (including a couple I didn’t ask), along with my original answers.

  • Best Named SQL feature: Hekaton was, at least until marketing got in the way and made it In-Memory OLTP.
  • Worst Named SQL feature: The timestamp data type.  As a runner-up, Azure Data Studio.
  • SQL Feature Most Likely To Have Been Named By Marketing: Always On Availability Groups.  Or maybe AlwaysOn.  I think it might have been Always-On at one point as well, but don’t recall.
  • Most Accurately Named SQL Feature: I want jokingly to say “Priority boost” here.  On the plus side, most features are named clearly enough, even if sometimes they take away nice names and give them marketing-heavy names.
  • Least Accurately Named SQL Feature: The timestamp data type.  As a runner-up, Azure Data Studio.
  • SQL Feature That Should Throw A Hard 24 Error If You Try to Use It In A New Project (But Continue To Work For All Your Existing Crap): Non-schemabound table-valued functions
  • SQL Feature That Just Needs A Little Love to Be Great: If this was 2010, I would have said Service Broker.  Right now, PolyBase.
  • SQL Feature That Can Never Be Great Even With Barry White Levels of Love: CLR strict security.
  • Best SQL Feature: Metadata is also described in terms of SQL.
  • Suckiest SQL Feature: Stretch DB is too easy.  CLR strict security could be on here, but I’ll pick something different:  SSIS scale-out.  It’s a half-hearted attempt to say that SSIS scales.
  • Surprisingly Useful SQL Feature: Batch mode processing.  To clarify, the surprising part is just how often batch mode processing can speed things up.
  • Surprisingly Useless SQL Feature: Temporal tables.  I think they tried to solve two problems at once and ended up providing a halfway-viable solution to each, but not a complete solution to either. A bit of that Barry White love could fix this.
  • SQL Feature I’ll Probably Never Use But I Like That It Exists: Query Store plan forcing for cursors.  It’s easy to dunk on cursors and performance, but I appreciate that they were able to figure this out.

Leave a Reply

Your email address will not be published. Required fields are marked *