Based on Technet, Query Governor
Cost Limit refers to the
estimated elapsed time, in seconds. This meets this excerpt from the Microsoft SQL Server 2008 Administrator's Pocket Consultant book.
Walking to the book Microsoft SQL Server 2012 Unleashed, pages 30 and 31, the
cost limit is
proportional to the estimated time, and one should apply a factor (K), to take into account the system's specifics, so it would be
K x time(s).
Now, if I take this (somehow older) article, it feels like establishing a relationship between cost limit and time is not doable.
Can I say that in the real world, despite some more rare exceptions, the 2nd approach could be taken as a rule of thumb?
Based on my notes from SQL Server internals (by Kalen Delaney).... below is my understanding
Query Governor Cost limit is the max no. of seconds that a query can run.
So, essentially :
Note that the value set for seconds isn’t clock-based. It corresponds to seconds on a specific hardware configuration used during product development, and the actual limit might be higher or lower on your machine.
Also, it should be noted that changing the
Query Governor Cost limit will flush the entire plan cache resulting in generating a new plan afterwards.