Re: LRU latches

From: Dave Haas <davidh_at_--nospam--hotmail.com>
Date: Sat, 24 Feb 2001 23:09:09 GMT
Message-ID: <paXl6.15430$4q3.1605146@news1.telusplanet.net>

Hi Thomas.

A latch is a memory lock. Since the SGA is the "SHARED Global Area", many processes can potentailly be writing to the log buffer at the same time (all of the server or "shadow" processes). Latches control both the LRU list for the buffer cache and the redo log buffer.

Therefore, the system needs to provide exclusive access to the log buffer to an individual process (or processes) and this is done with a latch (it's essentially a sephamore).

So, that being said, the parameters are:

LOG_SIMULTANEOUS_COPIES : # of latches controlling the log buffer. LOG_SMALL_ENTRY_MAX_SIZE : the size of a redo log entry UNDER which a server process will simply use the redo ALLOCATION latch for the copy into the log buffer instead of attempting to obtain an real redo COPY latch.

HTH, Dave Haas

"Thomas Tomello" <tom_at_tatsoft.com> wrote in message news:tPRl6.3253$Ea1.219013_at_bgtnsc05-news.ops.worldnet.att.net...
> Could someone explain LRU latches and the relationship of these to the
> LOG_SIMULTANEOUS_COPIES and LOG_SMALL_ENTRY_MAX_SIZE parms? Thanks...
>
>
Received on Sat Feb 24 2001 - 17:09:09 CST