This may seems like a stupid question, but I cant seem to find a good answer for it. I have a pretty good intuitive understanding of what
ROWS_READ in for example
sysibmadm.snapdyn_sql means, and what conclusions can be drawn from this metric. However, I'm not sure exactly what this means, and I can't find any information in the documentation. Looking at:
The number of rows read from the table.
My question is due to observations that sometimes ROWS_READ does not increase eventhough NUM_EXECUTION does. At the same time there are situations where ROWS_READ increases along with NUM_EXECUTIONS eventhough data definitely fits in bufferpool.
Are ROWS_READ physical reads (from disk to bufferpool) or logical reads (in bufferpool) or orthogonal to physical/logical? If the metric is measured in the bufferpool, how can the first phenomena be explained? Invalidated package (something wrong with counters in package cache), a sudden change to index only access or something else?