Page Cleaning of Buffer Pools

[Pierre Saint-Jacques]

DB2 V8.2 has a new envir. var.
DB2_USE_ALTERNATE_PAGE_CLEANING=YES
The docs. mention that this will make DB2 ignore chngpgs_thresh and use
softmax to even the rate of writing out of the bp's.

What agent process then takes over.

On a workload of 10000 trans./min., I get over 9000 bp's writes out of
my snapshot. However, LSAN Gap Cleaners, Thereshold Cleansers and
Victim Page Cleaners are all = 0

So what does the cleaning???
Thanks, Pierre.
--
Pierre Saint-Jacques - Reply to: sescons at attglobal dot net
IBM DB2 Cerified Solutions Expert - Administration
SES Consultants Inc.

[Mark A]

"Pierre Saint-Jacques" <se*****@attglobal.net> wrote in message
news:Ce**********************@wagner.videotron.net ...

DB2 V8.2 has a new envir. var.
DB2_USE_ALTERNATE_PAGE_CLEANING=YES
The docs. mention that this will make DB2 ignore chngpgs_thresh and use
softmax to even the rate of writing out of the bp's.

What agent process then takes over.

On a workload of 10000 trans./min., I get over 9000 bp's writes out of
my snapshot. However, LSAN Gap Cleaners, Thereshold Cleansers and
Victim Page Cleaners are all = 0

So what does the cleaning???
Thanks, Pierre.
--
Pierre Saint-Jacques - Reply to: sescons at attglobal dot net

My understand is that DB2_USE_ALTERNATE_PAGE_CLEANING=YES uses a more
aggressive page cleaning algorithm that in your case starts page cleaning
before the LSAN Gap Cleaners, Threshold Cleansers, and Victim Page Cleaners
are triggered by their own algorithms.

[Josh Tiefenbach]

Pierre Saint-Jacques wrote:

On a workload of 10000 trans./min., I get over 9000 bp's writes out of my snapshot. However, LSAN Gap Cleaners, Thereshold Cleansers and
Victim Page Cleaners are all = 0

So what does the cleaning???

The documentation for LSN Gap/threshold/dirty steal triggers state that
when DB2_USE_ALTERNATE_PAGE_CLEANING is set to on, those elements are
always reported as 0.

As Mark points out, a completely different algorithm is used to
determine when to start writing out pages, and which pages to write
out. Page cleaners no longer respond to the 'traditional' triggers (ie,
those expressed in the monitor elements you quote).

jsoh

[Pierre Saint-Jacques]

I had missed that in my search.
Thanks all for the feedback, Pierre.

Josh Tiefenbach wrote:

Pierre Saint-Jacques wrote:

On a workload of 10000 trans./min., I get over 9000 bp's writes out

of

my snapshot. However, LSAN Gap Cleaners, Thereshold Cleansers and
Victim Page Cleaners are all = 0

So what does the cleaning???

The documentation for LSN Gap/threshold/dirty steal triggers state that
when DB2_USE_ALTERNATE_PAGE_CLEANING is set to on, those elements are
always reported as 0.

As Mark points out, a completely different algorithm is used to
determine when to start writing out pages, and which pages to write
out. Page cleaners no longer respond to the 'traditional' triggers (ie,
those expressed in the monitor elements you quote).

jsoh

--
Pierre Saint-Jacques - Reply to: sescons at attglobal dot net
IBM DB2 Cerified Solutions Expert - Administration
SES Consultants Inc.