Re: DBA_HIST_FILESTATXS information

Freek,

If you check v$sqlstats (or any v$sql% view), you'll see a query like this:

insert into wrh$_filestatxs (...fw.count, fw.time from x$kcfio fs, file$ df, x$kcbfwait fw where...

The wrh$_filestatxs table is in the definition of DBA_HIST_FILESTATXS. So we know the wait_count and time come from x$kcbfwait. My notes about this x$ table says:

A commonly used query breaks down the contents of v$waitstat into per-datafile statistics: select name, count, time from v$datafile df, x$kcbfwait fw where fw.indx+1 = df.file#

I think it's safe to say the concept of wait here is different from that in the events "db file sequential/scattered read". Here it refers to buffer busy wait (or other names in 10g), for instance, when a session tries to read a datafile block into buffer cache while another session is already doing so. Because of this difference, we can't compare the wait count and time here with the other columns. Oracle provides these two columns just for our convenience.

Yong Huang

> Does anyone has some more information about the DBA_HIST_FILESTATXS view?
> I’m trying to understand the difference of the read_time/write_time columns
> and the time column.
>
> According to the reference guide, the time column holds the wait time, but
> which waits?
> It is my understanding that all reads list their time in the “db file
> sequential / scattered read” wait events, but the value in the time column
> seems to be always smaller then the values in the readtime column.
>
> Also, are all columns cumulative or only the SINGLEBLKRDTIM column?
>
>
> Regards,
>
>
> Freek D'Hooge
      

--
http://www.freelists.org/webpage/oracle-l