The morning after the Data TLV Summit in Tel Aviv, I stood out on the balcony and answered a few of your questions from https://pollgab.com/room/brento, rapid-fire style:
Here’s what we covered:
- 00:00 Intros
- 00:47 cyrpl: Hi Brent from PEI, Canada. A support person told me today that when creating a table, an ID column should almost be a default for every table. Do you agree? 98% of our tables do not have ID fields. Would adding ID fields to busy tables generally help query performance?
- 01:15 Nas: What is quickest and short downtime way of migrating 7TB database on new storage old SAN (DELL) new SAN(HPE)?
- 02:19 Simon Frazer: Have you come across scenarios where you feel SQL Server would’ve benefited from allowing more than 1 8k page for statistics?
- 02:38 Ingibjorg: Do Postgres DBAs tend to have higher pay ceilings than MSQL DBAs?
- 03:12 Ingibjorg: What tools / tests do you like to use when performance testing the network layer for a new SQL server?
- 03:32 sp_blitz_says_myboss_is_a_rocket_scientist: in what cases should we create statistics ourselves?
- 04:33 Does Basically Anything: Hi Brent! How would you recommend tuning when faced with a cursor query that has so many iterations that you can’t pull an actual execution plan when running it because it overwhelms SSMS?
- 05:06 Ingibjorg: Management uses KPI’s to judge the developers. Management also wants KPI’s to evaluate the SQL production DBA. In addition to meeting RPO/RTO objectives, what are good / bad KPI’s for this? Tickets closed, peer review scores, etc?
- 05:47 Mehmet: At what point should you ask dev to introduce app layer caching for a query that is rapidly re-executed?