Re: number of redolog groups

"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<3faebf7c$0$3792$afc38c87_at_news.optusnet.com.au>...
> "Sybrand Bakker" <gooiditweg_at_sybrandb.nospam.demon.nl> wrote in message
> news:8dctqv0nec0g2vjfa09kdgf67r6g7dmsd2_at_4ax.com...
> > On 9 Nov 2003 11:22:08 -0800, drak0nian_at_yahoo.com (Paul Drake) wrote:
> >
> > >As Oracle can use all available redo log members to feed the archiver
> > >processes, having more than 2 members per group could help to speed up
> > >archiving.
> >
> > I seem to recall someone in this group (HJR?) emphatically denied the
> > archiver would use multiple online redo logs to construct an archive.
> > Surely someone must not be speaking the truth.
>
> I don't believe I would have said something like that (but I suppose
> anything is possible). I think I am on record as saying that multiple ARCH
> processes do not read against a single group (ie, construct an archive log
> in parallel) -maybe that's what you're thinking of?
>
> The toppic has just been discussed on the Oak Table, and the consensus
> appeared to be that striped reads across multiple members by a single ARCH
> for a single group was definitely what happened in 8.0, but that with the
> appearance of multiple mandatory and optional destinations in 8i, the
> situation could be more complex. On the other hand, those mandatories and
> optionals only kick in with the Enterprise Edition, so the 8.0 behaviour is
> *most probably* the same in the higher versions if you are using the
> log_archive_dest and log_archive_duplex_dest parameters True for most Oracle
> installations on the planet). But I admit to not having tested it lately.
>
> Regards
> HJR
I would be interested to know this. Perhaps someone with access to truss on Solaris or tusc on HP can follow the arch process and verify this?

Regards,

Steve Received on Mon Nov 10 2003 - 12:27:41 CST