Re: SAN revisited

(my) practice behind it :

I was checking the backup and find corrupt Archives. First thought of oracle bug as we run 8.1.7.0, but then pulled one Set of Archives to lokal disk and did a filecompare (widow2000 adv. Server)

C:\Oracle\oradata\arcmach11>comp O:\arcmach12\* * Comparing O:\arcmach12\MACHP1T001S826.ARC and MACHP1T001S826.ARC... Files compare OK

Comparing O:\arcmach12\MACHP1T001S827.ARC and MACHP1T001S827.ARC... Files compare OK

.
.
.

Comparing O:\arcmach12\MACHP1T002S848.ARC and MACHP1T002S848.ARC... Files compare OK

Compare more files (Y/N) ? n

C:\Oracle\oradata\arcmach11>comp O:\arcmach12\* * Comparing O:\arcmach12\MACHP1T001S826.ARC and MACHP1T001S826.ARC... Files compare OK

Comparing O:\arcmach12\MACHP1T001S827.ARC and MACHP1T001S827.ARC... Files compare OK

.
.
.

Comparing O:\arcmach12\MACHP1T002S849.ARC and MACHP1T002S849.ARC... Files compare OK

Compare more files (Y/N) ? n

C:\Oracle\oradata\arcmach11>comp O:\arcmach12\* * Comparing O:\arcmach12\MACHP1T001S826.ARC and MACHP1T001S826.ARC... Files compare OK

Comparing O:\arcmach12\MACHP1T001S827.ARC and MACHP1T001S827.ARC... Compare error at OFFSET 6EA000
file1 = 49
file2 = 3B
Compare error at OFFSET 6EA001

o:\ ... SAN DISK
C:\ .. LOCAL DISK Funny isnīt it ?

regards

Stefan kapitza

tilwenbr_at_netscape.net (Mike Borden) wrote in message news:<8a41a7df.0206060617.2aecf7e5_at_posting.google.com>...
> Ok, I'm going to tell you what we did and the theory behind it. Also,
> I haven't read the white paper that is previously posted, but I will
> now.
>
> We put the redo logs and archived redo's on the local disk. We have an
> IBM AIX box using ssa, and the SAN is an IBM Shark. The theory behind
> this is, if the Shark was writting bad cache to the redo logs, the DB
> couldn't be recovered. Yes, the same thing could happen on the local
> box. However, if the SAN was writting bad cache to the DB files and
> the redo logs weren't getting the bad cache writes, cause they are on
> the local disks. The database could be recoverd. If everything was on
> the SAN then the DB would be trashed for sure. You would have to pull
> out backups before the SAN starting writting the bad cache to disks.
> Basically this is making sure that there wasn't a single point of
> failure. Like I said that was the theory here.
>
> So far, we haven't had any bottle necks by putting the redo logs and
> archived redo logs on the local SSA drives. Yes they are mirrored.
>
> Actually, I'm going to read the white paper now to see what it says.
>
>
>
>
> skapitza_at_volcanomail.com (s.kapitza) wrote in message news:<26703915.0206052247.7fc1647a_at_posting.google.com>...
> > hi,
> >
> > i m having here a compaq san.
> >
> > my expirience is, that you should have at least a mirror of the
> > archives (if you have one) on the local disk. So if your San is going
> > to sleep, you could
> > probably recover at least your database somewhere else.
> >
> > regards
> >
> > stefan kapitza
> >
> >
> > spamdump_at_nospam.noway.nohow (Ed Stevens) wrote in message news:<3cfe711b.88492845_at_ausnews.austin.ibm.com>...
> > > A few weeks ago I posted, requesting general comments and experiences on the use
> > > of SAN with Oracle databases. Now I have a followup.
> > >
> > > Given a server connected to a SAN (this will be Dell hardware if it makes any
> > > difference) and hosting multiple databases, is there any performance or
> > > administrative advantage to keeping some of the files on local drives?
> > >
> > > What I'm thinking here is that perhaps local drive access will be somewhat
> > > faster than the SAN and so might be a better location for rollbacks and temp
> > > space.
Received on Fri Jun 07 2002 - 03:41:34 CDT