On Apr 26, 2007, at 4:02 AM, Salak Juraj wrote:
Hello Richard!
I understand your arguments are "either veracity or performance".
I believe the point you possibly oversee is that
if db reads were made from both sides of the mirror
then db backup performance would (with high probabilty)
significantly rise
without harming database veracity.
However, this is theory only, as this is apparently not implemented
in TSM DB, point.
Consequently both your and Orville´s conclusions are correct.
Combining reliability and speed in current designs would lead to
implementig both HW mirroring and TSM mirroring, at extra costs$.
Hi, Juraj -
Oh, I definitely appreciate the potential gains possible if all
mirrors were employed for reads. (In the disks world, "the more
arms, the better".) But, as we know, that is not TSM's current
methodology.
There is the possibility of this situation changing, however, as
pondered in paper
"A Close-up Look at Potential Future Enhancements"
available at http://tsm-symposium.oucs.ox.ac.uk/
2005contributions.html .
Perhaps this year's Oxford Symposium will air further momentum in
this direction.
Richard Sims
