The challenge was to create a baseline from Oracle database viewpoint which allow us to spot after a storage migration if the performance changed.
I created a “magical number” called Whole Mbytes per seconds. Which is the total write and read of the whole database operations, including background processes, redos write and of course the foreground processes divided by the IO wait per second.
select total_mb/trunc(value/100) from v$sysstat,
(select trunc(sum(value)/1024/1024) total_mb from v$sysstat where name in
(‘physical read total bytes’,’physical write total bytes’))
where name=’user I/O wait time’;
Leave a Reply