Re: Redo Log Buffer Latches

Oracle uses the redo copy latch when the size of the log entry is smaller then log_small_entry_maxsize. The redo copy latch is hold for the whole time the log entry needs for the copy into the log buffer. The allocation latch is use only for allocate the room in the log buffer.

If there are really problems with the copy latch you can increase the parameter log_simultaneous_copies, which is DEFAULT 0. Another way is to set log_small_entry_max_size.

In article <6k1ic5$jj6$1_at_blackice.winternet.com>,   "Mike Quist" <mike.quist_at_supersolution.com> wrote:
>
>
> Vincent Birlouez wrote in message <356068D4.1C815CC_at_hotmail.com>...
> >
> >
> >The Oracle tuning guide says that the ratio MISSES/GETS for redo copy
> >should be less than 1%...As you can see below, my ratio is about 100% !
> >In the other hand, there is only 28 gets for a whole day. I had the
> >feeling that it's not really a problem. What do you think ?
> >
> >
> >NAME GETS MISSES Ratio <1%
> >IMMEDIATE_GETS IMMEDIATE_MISSES Ratio <1%
> >---------------------------------------- --------- --------- ---------
> >-------------- ---------------- ---------
> >redo allocation 537571 369
> >.0686421 0 0 0
> >redo copy 28 28
> >100 395943 184 .04644963
> >
> >Thanks.
> >
> Vincent,
> I don't see a problem here since the nowait miss ratio is OK, and
> that is the normal mode for the copy latch acquisition.
>
> But... the question I've had for some time now is:
> When does Oracle make a willing-to-wait request for the redo
> copy latch? Any guru's out there that know the answer to this?
> Thanks,
> Mike
>
>

-----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading Received on Fri May 22 1998 - 02:12:12 CDT