When you say PBM was correct but too early, I'd qualify that SQL Server 2008's idea of PBM was correct, but the implementation certainly was not. The checks were done post-execution - like, after the index was already built, the whole thing would be rolled back for having the wrong name.
Plus the real things we needed PBM for in the field (like backups and corruption checking) didn't ship. I really don't care what the index's name is - I care that it's backed up and protected. Microsoft went after the low-hanging fruit with PBM, but that fruit was low-hanging because nobody wanted to eat it.
Thanks Brent - fair point on the implementation and scope. I could very much see the potential though and had hoped (at the time as a PFE) that it would expand. Now I feel like the broader vision is needed more than ever - before we have an agentic-driven database-a-slop-alypse.
When you say PBM was correct but too early, I'd qualify that SQL Server 2008's idea of PBM was correct, but the implementation certainly was not. The checks were done post-execution - like, after the index was already built, the whole thing would be rolled back for having the wrong name.
Plus the real things we needed PBM for in the field (like backups and corruption checking) didn't ship. I really don't care what the index's name is - I care that it's backed up and protected. Microsoft went after the low-hanging fruit with PBM, but that fruit was low-hanging because nobody wanted to eat it.
Thanks Brent - fair point on the implementation and scope. I could very much see the potential though and had hoped (at the time as a PFE) that it would expand. Now I feel like the broader vision is needed more than ever - before we have an agentic-driven database-a-slop-alypse.