FAQ
An Oracle Support technician recommended that I try to aim for three log
switches per hour, during busy times.

According to the throughput I got yesterday between 3 and 4 o'clock PM, my
redo log files would each have to be 1G in size.

That seems much too large to me, and I am not even sure it is possible to
create redo log files that size. Under Oracle 7.3. the largest I could
create on Tru64 UNIX 4.0g was 40M redo log files.

Is this another "rule of thumb" that should go down the drain?

Regards,
Patrice Boivin
Systems Analyst (Oracle Certified DBA)

Systems Admin & Operations | Admin. et Exploit. des systèmes
Technology Services | Services technologiques
Informatics Branch | Direction de l'informatique
Maritimes Region, DFO | Région des Maritimes, MPO

E-Mail: boivinp_at_mar.dfo-mpo.gc.ca
--

Please see the official ORACLE-L FAQ: http://www.orafaq.com
--

Author: Boivin, Patrice J
INET: BoivinP_at_mar.dfo-mpo.gc.ca

Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
San Diego, California -- Public Internet access / Mailing Lists
--------------------------------------------------------------------

To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from). You may
also send the HELP command for other information (like subscribing).

Search Discussions

  • Khedr, Waleed at Mar 14, 2002 at 3:03 pm
    In my opinion not less than 5-10 minutes in very busy times.

    -----Original Message-----
    Sent: Thursday, March 14, 2002 8:19 AM
    To: Multiple recipients of list ORACLE-L

    An Oracle Support technician recommended that I try to aim for three log
    switches per hour, during busy times.

    According to the throughput I got yesterday between 3 and 4 o'clock PM, my
    redo log files would each have to be 1G in size.

    That seems much too large to me, and I am not even sure it is possible to
    create redo log files that size. Under Oracle 7.3. the largest I could
    create on Tru64 UNIX 4.0g was 40M redo log files.

    Is this another "rule of thumb" that should go down the drain?

    Regards,
    Patrice Boivin
    Systems Analyst (Oracle Certified DBA)

    Systems Admin & Operations | Admin. et Exploit. des systèmes
    Technology Services | Services technologiques
    Informatics Branch | Direction de l'informatique
    Maritimes Region, DFO | Région des Maritimes, MPO

    E-Mail: boivinp_at_mar.dfo-mpo.gc.ca
    --

    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --

    Author: Boivin, Patrice J
    INET: BoivinP_at_mar.dfo-mpo.gc.ca

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------

    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
    --

    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --

    Author: Khedr, Waleed
    INET: Waleed.Khedr_at_FMR.COM

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------

    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
  • Post, Ethan at Mar 14, 2002 at 3:49 pm
    Nothing wrong with 1 GB or ever larger redo log files. Search Usenet at
    google for what "HJR" and others have to say on the subject. Howard Rogers
    (HJR) recommends one switch per day (manually) if possible and not more than
    once per hour. This is all to limit when a checkpoint actually occurs. Of
    course this means you better make sure you have your redo logs mirrored and
    multiplexed and instance recovery could be significant.

    Ethan

    -----Original Message-----
    Sent: Thursday, March 14, 2002 9:04 AM
    To: Multiple recipients of list ORACLE-L

    In my opinion not less than 5-10 minutes in very busy times.

    -----Original Message-----
    Sent: Thursday, March 14, 2002 8:19 AM
    To: Multiple recipients of list ORACLE-L

    An Oracle Support technician recommended that I try to aim for three log
    switches per hour, during busy times.

    According to the throughput I got yesterday between 3 and 4 o'clock PM, my
    redo log files would each have to be 1G in size.

    That seems much too large to me, and I am not even sure it is possible to
    create redo log files that size. Under Oracle 7.3. the largest I could
    create on Tru64 UNIX 4.0g was 40M redo log files.

    Is this another "rule of thumb" that should go down the drain?

    Regards,
    Patrice Boivin
    Systems Analyst (Oracle Certified DBA)

    Systems Admin & Operations | Admin. et Exploit. des systèmes
    Technology Services | Services technologiques
    Informatics Branch | Direction de l'informatique
    Maritimes Region, DFO | Région des Maritimes, MPO

    E-Mail: boivinp_at_mar.dfo-mpo.gc.ca

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------

    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may

    also send the HELP command for other information (like subscribing).

    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Post, Ethan
    INET: Ethan.Post_at_ps.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may

    also send the HELP command for other information (like subscribing).
  • Jeremiah Wilton at Mar 14, 2002 at 4:18 pm

    On Thu, 14 Mar 2002, Khedr, Waleed wrote:

    In my opinion not less than 5-10 minutes in very busy times.
    The three log switches per hour reccomendation from Oracle support is
    delusional. I would need 5 to 10 Gb logfiles to accomodate that
    reccomendation. And any standby would be miserably behind at all
    times. And recovery time (assuming log switch checkpoints) on
    instance failure would be unacceptable.

    Wall clock time is a bad metric for this question, and support should
    not have told Patrice this.

    If checkpoints are usually able to finish before the next log switch,
    then the logfile sizes are fine.

    Robert F., Here's one for ya.

    --
    Jeremiah Wilton
    http://www.speakeasy.net/~jwilton
    -----Original Message-----
    Sent: Thursday, March 14, 2002 8:19 AM
    To: Multiple recipients of list ORACLE-L

    An Oracle Support technician recommended that I try to aim for three log
    switches per hour, during busy times.

    According to the throughput I got yesterday between 3 and 4 o'clock PM, my
    redo log files would each have to be 1G in size.

    That seems much too large to me, and I am not even sure it is possible to
    create redo log files that size. Under Oracle 7.3. the largest I could
    create on Tru64 UNIX 4.0g was 40M redo log files.

    Is this another "rule of thumb" that should go down the drain?
    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Jeremiah Wilton
    INET: jwilton_at_speakeasy.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
  • Post, Ethan at Mar 14, 2002 at 4:49 pm
    Jeremiah,

    You are in a bit of an unusual situation, most databases are not producing 5
    to 10 Gb every 20 minutes. I grant the facts regarding instance recovery
    time and standby's. These should certainly be taken into consideration.
    However, it does seem frequent checkpoints are going to add overhead and
    that reducing the number of checkpoints which occur during periods of high
    activity would be beneficial. I usually go with large redo logs (as large
    as possible) and control me checkpointing with parameters if I have an SLA
    to meet. So I don't think clock time is completely out of the question.

    Ethan

    -----Original Message-----
    Sent: Thursday, March 14, 2002 10:19 AM
    To: Multiple recipients of list ORACLE-L
    On Thu, 14 Mar 2002, Khedr, Waleed wrote:

    In my opinion not less than 5-10 minutes in very busy times.
    The three log switches per hour reccomendation from Oracle support is
    delusional. I would need 5 to 10 Gb logfiles to accomodate that
    reccomendation. And any standby would be miserably behind at all
    times. And recovery time (assuming log switch checkpoints) on
    instance failure would be unacceptable.

    Wall clock time is a bad metric for this question, and support should
    not have told Patrice this.

    If checkpoints are usually able to finish before the next log switch,
    then the logfile sizes are fine.

    Robert F., Here's one for ya.

    --
    Jeremiah Wilton
    http://www.speakeasy.net/~jwilton

    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Post, Ethan
    INET: Ethan.Post_at_ps.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
  • Jeremiah Wilton at Mar 14, 2002 at 4:58 pm

    On Thu, 14 Mar 2002, Post, Ethan wrote:

    (HJR) recommends one switch per day (manually) if possible and not
    more than once per hour. This is all to limit when a checkpoint
    actually occurs.
    Ok, this must by a troll. Once a day? Manually? Is he kidding? A
    checkpoint is a normal thing. It is minimally impactful, and takes up
    a very small proportion of the DBW0 write batch. Real sites
    checkpoint all the time under heavy usage conditions and nobody gets
    hurt.

    --
    Jeremiah Wilton
    http://www.speakeasy.net/~jwilton
    -----Original Message-----
    Sent: Thursday, March 14, 2002 8:19 AM
    To: Multiple recipients of list ORACLE-L

    An Oracle Support technician recommended that I try to aim for three log
    switches per hour, during busy times.

    According to the throughput I got yesterday between 3 and 4 o'clock PM, my
    redo log files would each have to be 1G in size.

    That seems much too large to me, and I am not even sure it is possible to
    create redo log files that size. Under Oracle 7.3. the largest I could
    create on Tru64 UNIX 4.0g was 40M redo log files.

    Is this another "rule of thumb" that should go down the drain?
    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Jeremiah Wilton
    INET: jwilton_at_speakeasy.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
  • Freeman, Robert at Mar 14, 2002 at 5:08 pm
    <>

    As a baseline, I look for log switches to occur from somewhere between 10
    minutes and 15 minutes on my higher transaction systems that must perform.
    Thus, I have log files that are quite big.. Then I monitor and tune
    accordingly.

    The exposure to data loss is limited in 8i+ with the incremental checkpoints
    (and my segregated online redo log files), so I don't worry about it. The
    requirement in these systems is response, number one. Granted, if you have a
    small database, but high transactional volume, then the idea of having log
    files that are potentially larger than your whole database,
    might be counter intuitive, but it *is* a performance issue (and if you have
    a database with that kind of footprint, I'd think it was even more important
    from a performance reason because you have a great deal of volatility going
    on!)

    <>
    9i Data Guard offers several different solutions that make this no longer an
    issue. Of course, replication would also offer a solution to this problem
    (but not one of my favorite ones).

    <>
    Well, generally one metric is bad for any question... :-)) However, my
    general experience says that rapid log switching in a high transactional
    environment generally leads to performance impacts. If you require high
    concurrency, sub-second response, then disk space needs to not be a
    constraint. Look at your waits and see if they justify larger redo log
    sizes. I think generally you will find that in high volume systems, they
    will.

    <>
    BUT... a log switch does much more than influence checkpoints. There is
    latching that occurs, it causes the ARCH process to fire up and start moving
    the online redo log to the archive log directory, etc....

    RF

    Robert G. Freeman - Oracle8i OCP
    Oracle DBA Technical Lead
    CSX Midtier Database Administration

    The Cigarette Smoking Man: Anyone who can appease a man's conscience can
    take his freedom away from him.

    -----Original Message-----
    Sent: Thursday, March 14, 2002 11:19 AM
    To: Multiple recipients of list ORACLE-L
    On Thu, 14 Mar 2002, Khedr, Waleed wrote:

    In my opinion not less than 5-10 minutes in very busy times.
    The three log switches per hour reccomendation from Oracle support is
    delusional. I would need 5 to 10 Gb logfiles to accomodate that
    reccomendation. And any standby would be miserably behind at all
    times. And recovery time (assuming log switch checkpoints) on
    instance failure would be unacceptable.

    Wall clock time is a bad metric for this question, and support should
    not have told Patrice this.

    If checkpoints are usually able to finish before the next log switch,
    then the logfile sizes are fine.

    Robert F., Here's one for ya.

    --
    Jeremiah Wilton
    http://www.speakeasy.net/~jwilton
    -----Original Message-----
    Sent: Thursday, March 14, 2002 8:19 AM
    To: Multiple recipients of list ORACLE-L

    An Oracle Support technician recommended that I try to aim for three log
    switches per hour, during busy times.

    According to the throughput I got yesterday between 3 and 4 o'clock PM, my
    redo log files would each have to be 1G in size.

    That seems much too large to me, and I am not even sure it is possible to
    create redo log files that size. Under Oracle 7.3. the largest I could
    create on Tru64 UNIX 4.0g was 40M redo log files.

    Is this another "rule of thumb" that should go down the drain?
    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Jeremiah Wilton
    INET: jwilton_at_speakeasy.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Freeman, Robert
    INET: Robert_Freeman_at_csx.com

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).
  • Post, Ethan at Mar 14, 2002 at 6:33 pm
    I don't see any reason it should be completely out of the question. Not
    sure if you follow Usenet but I do think Howard has demonstrated himself to
    be pretty knowledgeable as you certainly have also. My personal "clock"
    time is to size my redo logs to switch no more than once per hour. Most
    databases are completely capable of this. I think in general checkpoints
    are costly. I have manually issued a checkpoint after large batch jobs and
    it definitely had a noticeable impact on cpu/disk and db performance. I have
    put some posts from Usenet from people who I think can be trusted for the
    most part. There is plenty more on this there. I say don't throw out clock
    time all together.

    Ethan

    CUT & PASTE from USENET BELOW FROM VARIOUS ***
    I wonder why many small log switches degrade performance? When there
    are more checkpoints, let's say every 15 minutes, the disk load will be
    distributed more evenly over the day in my opinion, or am I missing
    something?
    Bear in mind that when you switch out of a log group, DBWR only flushes
    dirty buffers covered by the log file being switched away from. Yet CKPT
    has to update the headers of *all* datafiles, and all controlfiles.

    Lots of small checkpoints are therefore grossly inefficient from the point
    of view of CKPT.

    Of course, your point has a degree of validity: the whole idea of
    fast_start_io_target is to get DBWR constantly dribbling dirtied blocks back
    down to disk -practically doing constant checkpointing.

    But it remains true that lots of small (or constant) checkpoints will result
    in even, but lower, performance. Better that, I suppose, than relatively
    high performance forever interrupted by mammoth checkpoints. But better
    still (in my opinion, anyway) is high performance uninterrupted by any
    checkpoints (at least until you manually force one at a time when no-one
    will care about the massive hiatus).

    Hence, I suppose, Oracle's hint that you'd only use fast_start_io_target
    when you absolutely, positively, have to bound your Instance Recovery time
    for the sake of an SLA.

    Regards
    HJR

    Connor McDonald wrote:
    I prefer the best of both worlds. I have "massive" redo logs (where
    "massive" means they virtually "never" switch by themselves) and then
    use dbms_job to manually switch them each 'n' minutes (where 'n' is
    typically 30 or 60).

    Then I know exactly when redo switches will be occurring; I know exactly
    how out of date my standby database will be; I can tweak the dbms_jobs
    to do it less frequently over night when large stuff is going on etc etc
    etc

    hth
    connor
    The smooth operation of the buffer cache will be disrupted for quite a
    while after a checkpoint.

    Cheers
    Nuno Souto
    nsouto_at_optushome.com.au.nospam

    You should switch at a rate that you're happy with.

    Sorry -that sounds like a bit of a non-answer, but it isn't. Log switches
    are good because they induce checkpoints. Once a checkpoint has been
    issued, you can pretty much guarantee that if you suddenly lost power on
    your machine and therefore suffered an Instance failure, transactions raised
    prior to the checkpoint will not need to be recovered. Therefore, frequent
    log switches bound Instance recovery time.

    On the other hand, log switches are also bad because they induce
    checkpoints. Checkpoints cause DBWR to flush buffers left, right, and
    centre, and also cause CKPT to have to do its thing to every data file and
    control file in the database. Masses of I/O mean bad performance. So
    frequent log switches cause woeful performance.

    Somewhere between trivially small Instance recovery times/dreadful
    performance (frequent switches)and huge Instance recovery times/good
    performance (few switches) is a point of equilibrium and compromise that
    you'll be able to live with. But it varies for each database, installation,
    and dba. You have to find your own happy compromise.

    On the other hand, lots of sites seem to think (as a sort-of rule of thumb)
    that a switch every half-hour or hour or so is acceptable.

    All you need then do is add log groups of a size that brings the rate of
    switching down to your chosen number, and drop the logs that are of a
    different size (because you can't resize existing logs, unfortunately).

    Regards
    HJR

    --
    Please see the official ORACLE-L FAQ: http://www.orafaq.com
    --
    Author: Post, Ethan
    INET: Ethan.Post_at_ps.net

    Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
    San Diego, California -- Public Internet access / Mailing Lists
    --------------------------------------------------------------------
    To REMOVE yourself from this mailing list, send an E-Mail message
    to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
    the message BODY, include a line containing: UNSUB ORACLE-L
    (or the name of mailing list you want to be removed from). You may
    also send the HELP command for other information (like subscribing).

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
grouporacle-l @
categoriesoracle
postedMar 14, '02 at 1:18p
activeMar 14, '02 at 6:33p
posts8
users5
websiteoracle.com

People

Translate

site design / logo © 2022 Grokbase