Robin McDonald wrote:
So basically you raise 3 excellent points
...
[snip]
2. ASM is a tool that runs on most OS oracle supports so to get the best
performance out of it you must do some tuning which is OS & storage
dependent. ASM is basically a "free" (once you've paid for your oracle
licence) clusterable extent based filesystem. So while you could say there
are other clusterable /  extent based filesystems (veritas & SAMFS) neither
are free.


My exposure to ASM was sufficiently traumatising that I decided
that if I ever worked for Oracle I would have to re-engineer it
so that it worked properly. (imnsho of course).

The main problem I encountered with it had to do with scsi 2 /
scsi 3 pgr reservations. As I recall it, ASM's method of reserving
a lun was to run "dd" in a loop since "dd" is a command which does
the same thing on all their target architecture.

This was the problem for me, is that dd does not actually set a
scsi2 pgre / scsi3 pgr key on the lun itself... and ASM basically
left itself a window of opportunity to die horribly in the short
period of time between one dd finishing and the next one starting.


I *hope* that ASM has changed since I had to deal with it.


James C. McPherson
(daddy make the bad ASM man stop!)
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to