What is considered to be a high cost?

Hi all

IM on oracle 10g r2 and Ive started to spend a lot more time to optimize (or start to :=-) questions.) My question is this: lets say that my query uses all my index, without full scans, buffer gets doesn't seem too bad, how do you know if the cost is still high? or rather if you can try to further reduce this number? for example, let's say that im a table query
with a predicate that uses 5 fields in the table, I have @180,000 total lines and I have a total cost of @250. I just wanted to get some thoughts out there, so that I can better understand my optimization strives... thx in advance - jp

I suggest the book based on the costs of Oracle Fundamentals, which explains clearly and in detail how to understand what is a high cost.

In particular, it is important to understand that the size of the cost is not so important as the relative cost of different plans. The cost is the idea of optimizers of how long it will take. So really, what you're looking for, this is where the optimizer is not doing things. Once, it was 'just' a matter of arithmetic of adding up different ways to access the lines - see the guide to performance on the nested loops, etc. - While in the more modern versions the optimizer can take into account the workloads, so listen to what you say on the histograms and the basic lines and so on, including using different places to keep the data in memory.

I think that people may be more willing to work through this with you, if you have a specific example and follow the instructions "how to publish an application for tuning. There are a lot of previous assignments like this here.

If you really want the details, look on the net for the 10053 trace. As http://tonguc.wordpress.com/2007/03/21/how-oracle-optimizer-may-decide-between-an-index-or-table-scan-access-path/

Tags: Database

Similar Questions

  • What is considered high Kbps for a virtual machine?

    I try to analyse some virtual machines to determine how they use disk latency. VMware performance dashboards I collect graphs showing Kbps on an individual basis of the VM. I see very significant differences on existing virtual machines.

    I have two virtual machines that have hit an average of 2000 to 300 Kbps but peaks up to 20000 and 3000.

    What is considered as high Kbps?

    I know that this manufacturer of disk and the type of Raid is a factor here. EMC tools monitor the latency of disk IO of reading and writing, and I do not know how to make the link between these in Kbps.

    Thanks for any info here.

    Mark

    Then, your best bet is to use ESXTOP to the command line and drill down through the HBA view.  First of all understand that vmhba is the LUN you want to watch and then explore the vmhba railways until you can get the world ID's of the virtual machine on the LUN.

    Unfortunately then you will have to play the game to figure out that VM is what ID World expanding views and... It's just a really annoying process.  I hope that you would not have to do.  I wish that VMware had a simple view of the LUN VM WITH name.  The drive see esxtop frankly sucks.

    If anyone knows a better way to achieve this, please chime.

  • I was wondering what is considered an "app"?

    Original title:

    OneDrive App sync issue on Windows 8

    I'm trying to do a factory reset on my lap top and wanted to synchronize some of my previous data after.

    When the sync 'Apps - the list of the apps I have installed' with Onedrive times connected to your Microsoft account, I was wondering what is considered an "app"? It of just something I bought in the app store or is this all I have previously installed/downloaded on my computer? For example, if I downloaded Diablo 3, which is an application or a file separated I should have saved on a USB? What I have to redownload all apps or with synchronization, it does it automatically?

    Thank you.

    Wednesday, June 3, 2015, 17:53:26 + 0000, gardeniagirl11 wrote:

    When the sync 'Apps - the list of the apps I have installed' with Onedrive times connected to your Microsoft account, I was wondering what is considered an "app"? It of just something I bought in the app store or is this all I have previously installed/downloaded on my computer?

    Unfortunately the term "app" has recently come to mean something of very
    specific to many people: a program that is called when you click on a
    tile in the subway/modern interface.

    But this isn't what this means for everyone. As far as I'm concerned,
    "app" is simply short for "application", and "the application is a Word.
    This basically means the same thing as the "program". So for me, all the
    the programs are "apps".

    That means Microsoft by "app" in this particular case, I have no idea.
    They have confused a lot of people with their non-traditional use of the
    Word and they are inconsistent in its use.

  • What process in the database is high in the Oracle system resources

    Hello everybady,

    I am beginner.

    How can we control or analize what process in the database is high in the Oracle database system resources?

    Could you please help this issue?

    Thanks and greetings



    A tuba.

    Please read SQL and PL/SQL FAQ

    ------------------
    Sybrand Bakker
    Senior Oracle DBA

  • What is the advantage of the high water line?

    Hello

    When I remove the enormous data in table then the high water mark always on the same position.

    1. then deleted block still on the data file then why? I understand that these blocks is not used and it is empty so when he used?

    2-why oracle do not recover these block when automatically delete us the records? and we must do to help shrink the command?

    3 - Why before retractable data we must allow movement of line? What is the meaning of movement of the line. and why it is disable by default, no impact on performance so afford him?

    4. i try to reduce and remove the space allocation for some gigantic data, but still tablespace free space is always the same, does that mean?

    Thank you

    (1) the deleted data are always in the data file, because it is the least expensive way to implement a delete.  It is much faster to just things like free brand rather than incur the cost of, say, block formatting when the last line is deleted.  Most of the time, you remove one or two lines - it would be unwise to undergo as a load whenever you delete a few lines.  Free blocks will be used during the insertion of subsequent data (or during subsequent updates increase the size of existing lines and have to migrate to the new blocks).

    (2) because the shrinkage of the segment is expensive.  If you did that every time that you deleted a single line, deletions would be orders of magnitude more expensive than they are today.

    (3) because the shrinkage of the segment means that you physically move potentially every row in the database.  From a functional point of view, this amounts to potentially remove all the ranks of the block in which it is and moving it to a different physical location at the 'head' of the segment so that all free space is at the "tail" of the segment and can be released.  Of course, it is unlikely that all ranks have actually be removed and reinserted.  But a significant fraction of them.  Among other things, this will change ROWID line that would break any code that was waiting for the ROWID remains constant, this is why you need to activate the movement of the line.

    (4) you must show us exactly what you are doing (the exact series of commands that you run, how you measure the free space, etc.).  If you just insert roughly the same amount of data in the table that you deleted over time, shrinkage of the segment is unlikely to be useful.  From the point of view of application performance, for example, you only care about high tide if you perform a full scan of the table.  But it is rarely wise to write code that is to do a full scan of a table that is large enough so that you can delete a 'huge' of data in the table number.

    Justin

  • What functions are part of the OLAP option, and what is considered "standard SQL"?

    Hello

    11.2 Enterprise Edition without OLAP license.

    Where can I find what functions are part of the OLAP option, which I'm not allowed to use without the extra cost of option?

    Not sure, what comes to my mind is such function as LISTAGG, CUBE, PARTITION BY and the like, but I would like a complete set so that we can ask our developers to avoid using functions that require an additional license.

    Thanks in advance

    None of the ones you mention are part of the OLAP option.

    The official license https://docs.oracle.com/cd/E11882_01/license.112/e47877.pdf document describes which includes the option.

    If you do not have him a license, then just make sure that it is not installed in the databases you create.

  • HP a1410n pavilloin... What memory... memory card high or low density?

    I need help... what should I buy, the high density or low density memory >? I know it's pc 3200.184 PIN... Thank you

    Bill,

    Go to Kiingston or Corsair and use the tools of memory. You can contact support technical and sales for a recommendation.

    If the memory is not compatible so why even I hear?

  • What is considered a slide view

    Hello

    Can someone tell me what need the requirements for a slide be counted as read in Captivate? I have the my data of Quiz results value and views of the slide and wonder what really means a slide view?

    Thank you.

    Hello

    If a blade is suspended by a button or click box, it depends on the associated action by clicking if the slide should be viewed for the entire timeline. If you choose the action "continue" the timeline will be visualized entirely, but if you choose action "Jump to slide" it will not be considered as a totally visualized slide. Hope you understand this explanation?

    For the last slide: if it is closed before the timeline was completely seen, it is not considered as read. It is very important, you need to force the user to click on a button, "Continuous" action attached to it.

    Lilybiri

  • What is behind low, medium or high LR_outputSharpeningLevel?

    What amount/RADIUS/details/hide values are used for each low, medium or high setting?

    TIA!

    Exit Lightroom sharpening uses a different algorithm than the standard accent. This algorithm takes into account that the media type, the output will be and the final resolution of the image. We worked with the Pixel Genius group to adjust these settings.

    So it is not all the equivalent amount, detail, RADIUS, masking values for these three levels; the algorithm that the pixels go through is quite different.

    -kevin

  • Oracle chooses a Plan with a higher cost?

    I have the following table with the lines > 870000
    SQL> desc t
     Name                                      Null?    Type
     ----------------------------------------- -------- ----------------------------
     DATE1                                              DATE
     STRING1                                            VARCHAR2(40)
     NUM1                                               NUMBER
     NUM2                                               NUMBER
     NUM3                                               NUMBER
     NUM4                                               NUMBER
     NUM5                                               NUMBER
     STRING2                                            VARCHAR2(3)
     NUM6                                               NUMBER
     STRING3                                            VARCHAR2(240)
     STRING4                                            VARCHAR2(240)
     STRING5                                            VARCHAR2(240)
     STRING6                                            VARCHAR2(240)
     STRING7                                            VARCHAR2(240)
     STRING8                                            VARCHAR2(240)
     STRING9                                            VARCHAR2(10)
    Create 2 indices
    SQL> CREATE INDEX t1 ON t (TRUNC("DATE1"), "NUM3");
    
    Index created.
    
    SQL> CREATE INDEX t2 ON t ("NUM3", "NUM1", TRUNC("DATE1")) ;
    
    Index created.
    Analyze all
    SQL> analyze index t1 compute statistics;
    
    Index analyzed.
    
    SQL> analyze index t2 compute statistics;
    
    Index analyzed.
    
    SQL> analyze table t compute statistics;
    
    Table analyzed.
    Oracle decides to use T2 to execute the query shown in the declaration of "implementation plan" below
      1  explain plan for
      2  SELECT
      3  NVL(SUM(num6),0) num6_SUM
      4  FROM t
      5  WHERE 1 = 1  AND num3 = :b1
      6  AND TRUNC(date1) BETWEEN sysdate-:b2 AND sysdate-:b3
      7* AND STRING9 = :b4
    SQL> /
    
    Explained.
    
    SQL> select * from table(dbms_xplan.display);
    
    PLAN_TABLE_OUTPUT
    --------------------------------------------------------------------------------
    Plan hash value: 1806095077
    
    --------------------------------------------------------------------------------------
    | Id  | Operation                     | Name | Rows  | Bytes | Cost (%CPU)| Time     |
    --------------------------------------------------------------------------------------
    
    PLAN_TABLE_OUTPUT
    -------------------------------------------------------------------------------------
    |   0 | SELECT STATEMENT              |      |     1 |    15 |   316   (4)| 00:00:04 |
    |   1 |  SORT AGGREGATE               |      |     1 |    15 |            |          |
    |*  2 |   FILTER                      |      |       |       |            |          |
    |*  3 |    TABLE ACCESS BY INDEX ROWID| T    |     2 |    30 |   316   (4)| 00:00:04 |
    |*  4 |     INDEX RANGE SCAN          | T2   |   356 |       |   270   (5)| 00:00:04 |
    --------------------------------------------------------------------------------------
    
    
    Predicate Information (identified by operation id):
    ---------------------------------------------------
    
    
    PLAN_TABLE_OUTPUT
    --------------------------------------------------------------------------------
       2 - filter(SYSDATE@!-TO_NUMBER(:B2)<=SYSDATE@!-TO_NUMBER(:B3))
       3 - filter("STRING9"=:B4)
       4 - access("NUM3"=TO_NUMBER(:B1) AND
                  TRUNC(INTERNAL_FUNCTION("DATE1"))>=SYSDATE@!-TO_NUMBER(:B2) AND
                  TRUNC(INTERNAL_FUNCTION("DATE1"))<=SYSDATE@!-TO_NUMBER(:B3))
           filter(TRUNC(INTERNAL_FUNCTION("DATE1"))>=SYSDATE@!-TO_NUMBER(:B2) AND
                  TRUNC(INTERNAL_FUNCTION("DATE1"))<=SYSDATE@!-TO_NUMBER(:B3))
    Yet, indeed, Oracle thinks the cost of T1 is smaller (it's also what I think)
      1  explain plan for
      2  SELECT /*+ index(t t1) */
      3  NVL(SUM(num6),0) num6_SUM
      4  FROM t
      5  WHERE 1 = 1  AND num3 = :b1
      6  AND TRUNC(date1) BETWEEN sysdate-:b2 AND sysdate-:b3
      7* AND STRING9 = :b4
      8  /
    
    Explained.
    
    SQL> select * from table(dbms_xplan.display);
    
    PLAN_TABLE_OUTPUT
    --------------------------------------------------------------------------------------
    Plan hash value: 1945175582
    --------------------------------------------------------------------------------------
    | Id  | Operation                     | Name | Rows  | Bytes | Cost (%CPU)| Time     |
    --------------------------------------------------------------------------------------
    
    
    PLAN_TABLE_OUTPUT
    --------------------------------------------------------------------------------------
    |   0 | SELECT STATEMENT              |      |     1 |    15 |    61   (0)| 00:00:01 |
    |   1 |  SORT AGGREGATE               |      |     1 |    15 |            |          |
    |*  2 |   FILTER                      |      |       |       |            |          |
    |*  3 |    TABLE ACCESS BY INDEX ROWID| T    |     2 |    30 |    61   (0)| 00:00:01 |
    |*  4 |     INDEX RANGE SCAN          | T1   |   356 |       |    14   (0)| 00:00:01 |
    --------------------------------------------------------------------------------------
    
    
    Predicate Information (identified by operation id):
    ---------------------------------------------------
    
    
    PLAN_TABLE_OUTPUT
    --------------------------------------------------------------------------------
       2 - filter(SYSDATE@!-TO_NUMBER(:B2)<=SYSDATE@!-TO_NUMBER(:B3))
       3 - filter("STRING9"=:B4)
       4 - access(TRUNC(INTERNAL_FUNCTION("DATE1"))>=SYSDATE@!-TO_NUMBER(:B2) AND
                  "NUM3"=TO_NUMBER(:B1) AND TRUNC(INTERNAL_FUNCTION("DATE1"))<=SYSDATE@!-TO_NUMBER(:B3))
           filter("NUM3"=TO_NUMBER(:B1))
    No idea why Oracle would use a plan which he thinks actually more expensive?

    Also, I know that EXPLAIN PLAN has some hypothesis about bind variable. So I also surrounded the dates with cast (xxx date), I even think that he can recognize sysdate as a date already. This has no effect on the plan and cost estimate.


    Oracle: 10.2.0.4, 9.2.0.6
    OS: RHEL 3

    Hello

    You collect statistics correctly.

  • Re: What is the Toshiba Recovery disk would cost?

    Thank you very much for your timely and valuable contributions.

    I tried the link below
    https://backupmedia.Toshiba.EU/landing.aspx

    and not able to get prices in Indain rupee.

    kindly can you (anyone) help me know,.
    How much does it cost in Indian rupee?

    Thanks in advance!

    I have not quite how it is safe, but I put m t think it s expensive. Maybe something like 30 or 40 Euros

  • What is the difference between ram high and low density?

    I have an ibm x 3455 7984. high density RAM will it work? If so, it will affect the speed for virtual machines?

    Hello

    If your system is designed to accept briefs DDR (Double Data Rate) DIMM 184pin (usually for desktop computers) or (usually laptops) SODIMM 200pin built for your system to recognize fully the capacity of 1 GB per slot of memory, you must use "Low Density - 64Mx8 config" module of 1 GB. If you use ' high-density - 128Mx4 config' 1 GB module, your PC can only recognize that it is HALF the size to 512 MB or most of the time it won't work.

    In summary: -.

    • The LOW DENSITY modules have 100% compatibility with all systems and all chipsets as it is entirely due to the JEDEC guidelines.
    • HIGH DENSITY modules only have 10% of compatibility and are VERY RIVER because it breaks any guidance Council JEDEC (Joint Electron devices Engineering)

    I also suggest you to contact the IBM support center and check.

    http://www-947.IBM.com/support/entry/portal/overview/software/WebSphere/WebSphere_Application_Server?lnk=mhsd00

    http://www.IBM.com/us/en/

  • What is the difference between the fixed cost, Costed and distributed?

    Hi friends,
    Can someone tell me the difference between a fixed cost, Costed and distributed, please?

    Thank you

    Check under the "Costing" title on the HRMS guide, the link is here
    http://docs.Oracle.com/CD/A60725_05/HTML/comnls/us/per/defelw04.htm
    Thank you

  • What after-sales repalcement batteries are high quality?

    I see a lot of different options, and I know that in the world of the iphone, there are a ton of crud is peddled. Yet, there is not that many manufacturers of batteries there, hgih quality units so many OEMs must be available.  Anyone know good? $130 on an older machine is a lot of money for a battery.  TIA,

    Grant

    Recommendation of Apple is that you use a Service Center authorized Apple Apple battery.

  • What is considered a trusted application

    Me and my brother made a soft and it's a chronometer, clock and room to launch application. We intend to update with new applications per week (depending on the approval time). My application is likely to get an approval?

    They want to increase their number of app and are unlikely to reject any application, unless they really have to reject. In 60 apps that I got an approval delayed because a screenshot showing a shell of BlackBerry and refused initially and then review extended more than a month after I made the changes they asked. All others were approved when they arrived around him.

    The app you describe seems like it would be approved as long as it works - there are less things already in the store.

    Good luck...

Maybe you are looking for