Skip to Main Content

Oracle Database Discussions

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

Increasing dbwr process to 1xCPU or 2x no of CPU cores will improve the system performance.

NanubalusSep 3 2014 — edited Sep 5 2014

Hello,

Greetings!!!

Currently we are having one DBWR with 4 dbwr_io_slaves & system is in NetApp Storage with 8 cpu's. I feel increasing dbwr process to 1xCPU or 2x no of CPU cores will improve the system performance. any idea?

Environment:

HP-UX

Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - 64bit

We have done following changes after NetApp implementation

disk_asynch_io=TRUE

filesystemio_options=SETALL

dbwr_io_slaves=4

ALTER SYSTEM RESET db_file_multiblock_read_count SCOPE=SPFILE SID='*';

Below are the dbwr statistics:

dbwr statistic                                 statistic value

---------------------------------------------- ---------------

DBWR checkpoint buffers written                     12,836,278

DBWR fusion writes                                           0

DBWR object drop buffers written                         1,763

DBWR parallel query checkpoint buffers written           4,036

DBWR revisited being-written buffer                          1

DBWR tablespace checkpoint buffers written                   0

DBWR thread checkpoint buffers written               1,753,757

DBWR transaction table writes                            6,581

DBWR undo block writes                               6,967,648

DEFAULT 8k buffer busy wait                          3,245,536

DEFAULT 8k free buffer wait                                  0

DEFAULT 8k hit ratio                                        97

DEFAULT 8k write complete                                    1

wait

buffer is not pinned count                       3,974,046,825

buffer is pinned count                          ##############

change write time                                      113,885

commit cleanout failures: buffer being written              40

commit cleanout failures: write disabled                     0

data block                                           3,239,864

dirty buffers inspected                              1,065,608

flashback log writes                                         0

free buffer inspected                              382,422,694

free buffer requested                              417,571,319

hot buffers moved to head of LRU                    46,874,546

lob writes                                           1,876,399

lob writes unaligned                                 1,876,208

no buffer to keep pinned count                             144

physical write IO requests                           6,202,118

physical write bytes                            ##############

physical write total IO requests                    15,565,428

physical write total bytes                      ##############

physical write total multi block requests            6,351,814

physical writes                                     37,146,489

physical writes direct                              22,049,905

physical writes direct (lob)                             3,642

physical writes direct temporary tablespace         21,861,535

physical writes from cache                          15,096,584

physical writes non checkpoint                      35,044,571

pinned buffers inspected                               549,802

summed dirty queue length                           11,856,553

switch current to new buffer                         1,462,350

write clones created for recovery                            0

write clones created in background                         157

write clones created in foreground                       1,905

Thank you,

Ram

Comments
Locked Post
New comments cannot be posted to this locked post.
Post Details
Locked on Oct 3 2014
Added on Sep 3 2014
14 comments
4,413 views