Poor Oracle Performance consistency running concurrent queries

Similar Questions

• Problems running SPARQL queries with Jena Oracle adapter

  Hello
  
  I am trying to run SPARQL queries using the adapter of Jena.
  
  I use Jena 2.6.0 and all concerned pots (granules and ojdbc5) and running on Oracle 11 g 1 material.
  
  I'm under theexample code provided with the adapter Jena:
  ' public void testOrderBy (String jdbcUrl, String user,
  String password, String modelName,
  PrintStream psOut)
  throws SQLException, IOException
  {
  Oracle Oracle = new Oracle (jdbcUrl, user, password);
  Chart GraphOracleSem = new GraphOracleSem (oracle, modelName);
  Model ModelOracleSem = new ModelOracleSem (graph);
  
  Void node = Node.createURI ("http://sub/a3");
  Pred node = Node.createURI ("http://pred/p1-a3");
  Node obj = Node.createLiteral("8888");
  Graph.Add (Triple.create (pred, sub, obj));
  
  void = Node.createURI ("http://sub/a2");
  PRED = Node.createURI ("http://pred/p2-a2");
  Graph.Add (Triple.create (pred, sub, obj));
  
  void = Node.createURI ("http://sub/a4");
  PRED = Node.createURI ("http://pred/p3-a4");
  Graph.Add (Triple.create (pred, sub, obj));
  
  Dim queryString As String = "SELECT? s? p? o"
  + "WHERE {?}" s? p? o.}"
  + "ORDER BY? s « ;
  
  psOut.println ("testOrderBy: question 1");
  runQuery (QueryFactory.create (queryString), model, psOut);
  
  
  String queryString2 = "SELECT? s? p? o"
  + "WHERE {?}" s? p? o.}"
  + "ORDER BY DESC (?) p) « ;
  
  psOut.println ("testOrderBy: request 2");
  runQuery (QueryFactory.create (queryString2), model, psOut);
  
  Graph.Close ();
  Oracle.Dispose ();
  }
  
  I'm getting the following exception:
  Exception in thread "main" java.lang.NoSuchMethodError: com.hp.hpl.jena.graph.query.Query.getTriples () mCol/hp/hpl/jena/graph/query/NamedTripleBunches;
  at oracle.spatial.rdf.client.jena.OracleSemQueryPlan.getStatements(OracleSemQueryPlan.java:363)
  at oracle.spatial.rdf.client.jena.OracleSemQueryPlan.executeBindings(OracleSemQueryPlan.java:182)
  to com.hp.hpl.jena.sparql.engine.iterator.QueryIterBlockTriplesQH$ StagePattern. < init > (QueryIterBlockTriplesQH.java:90)
  at com.hp.hpl.jena.sparql.engine.iterator.QueryIterBlockTriplesQH.nextStage(QueryIterBlockTriplesQH.java:56)
  at com.hp.hpl.jena.sparql.engine.iterator.QueryIterRepeatApply.makeNextStage(QueryIterRepeatApply.java:87)
  at com.hp.hpl.jena.sparql.engine.iterator.QueryIterRepeatApply.hasNextBinding(QueryIterRepeatApply.java:49)
  at com.hp.hpl.jena.sparql.engine.iterator.QueryIteratorBase.hasNext(QueryIteratorBase.java:71)
  at com.hp.hpl.jena.sparql.engine.iterator.QueryIterSort.sort(QueryIterSort.java:44)
  to com.hp.hpl.jena.sparql.engine.iterator.QueryIterSort. < init > (QueryIterSort.java:36)
  to com.hp.hpl.jena.sparql.engine.iterator.QueryIterSort. < init > (QueryIterSort.java:31)
  at com.hp.hpl.jena.sparql.engine.main.OpCompiler.compile(OpCompiler.java:292)
  at com.hp.hpl.jena.sparql.engine.main.CompilerDispatch.visit(CompilerDispatch.java:142)
  at com.hp.hpl.jena.sparql.algebra.op.OpOrder.visit(OpOrder.java:28)
  at com.hp.hpl.jena.sparql.engine.main.CompilerDispatch.compile(CompilerDispatch.java:33)
  at com.hp.hpl.jena.sparql.engine.main.OpCompiler.compileOp(OpCompiler.java:71)
  at com.hp.hpl.jena.sparql.engine.main.OpCompiler.compile(OpCompiler.java:298)
  at com.hp.hpl.jena.sparql.engine.main.CompilerDispatch.visit(CompilerDispatch.java:149)
  at com.hp.hpl.jena.sparql.algebra.op.OpProject.visit(OpProject.java:47)
  at com.hp.hpl.jena.sparql.engine.main.CompilerDispatch.compile(CompilerDispatch.java:33)
  at com.hp.hpl.jena.sparql.engine.main.OpCompiler.compileOp(OpCompiler.java:71)
  at com.hp.hpl.jena.sparql.engine.main.OpCompiler.compile(OpCompiler.java:49)
  at com.hp.hpl.jena.sparql.engine.main.QueryEngineMain.eval(QueryEngineMain.java:43)
  at com.hp.hpl.jena.sparql.engine.QueryEngineBase.createPlan(QueryEngineBase.java:81)
  at com.hp.hpl.jena.sparql.engine.QueryEngineBase.getPlan(QueryEngineBase.java:70)
  to com.hp.hpl.jena.sparql.engine.main.QueryEngineMain$ 1.create(QueryEngineMain.java:59)
  at com.hp.hpl.jena.sparql.engine.QueryExecutionBase.getPlan(QueryExecutionBase.java:262)
  at com.hp.hpl.jena.sparql.engine.QueryExecutionBase.startQueryIterator(QueryExecutionBase.java:239)
  at com.hp.hpl.jena.sparql.engine.QueryExecutionBase.execResultSet(QueryExecutionBase.java:244)
  at com.hp.hpl.jena.sparql.engine.QueryExecutionBase.execSelect(QueryExecutionBase.java:77)
  in the example. Examples.runQuery (Examples.java:1748)
  in the example. Examples.testOrderBy (Examples.java:1035)
  in the example. Test.main (test.) Java:23)
  
  Someone has an idea what is the problem? Am I guilty of?
  
  Doron,

  There was some interface with Jena itself changes.
  As a result, the Oracle Jena 2.0 adapter does not work with Jena 2.5.7 or Jena 2.6.0. Please try 2.5.6.
  We are working on a new version of the adapter of Jena and this restriction has been lifted.

  See you soon,.

  Zhe Wu

• How to suggest a new symbol of certification: (as OCP *), which includes certified more than professional oracle performance and tuning?

  Hello

  There are several versions Oracle has removed the requirement of performance and tuning in what concerns OCP.

  You can get a separate ECO - Oracle Certified Expert for the performance and tuning.

  However, databases there include performance and tuning in the basic DBA certification.

  I understand there is an OCP (for DBA) and a separate OCÉ (performance and tuning),

  but could it be considered as an additional symbol combining the OCP and OCE (such as OCP * or OCPE or OCP2).

  We could always get and OCP for DBA and a separate ECA for the performance and tuning.

  If there was an an additional symbol, then maybe selftest software would also create reviews of preparation for the performance and tuning.

  Perhaps more people write books on their performance and tuning of certification.

  Maybe Oracle would provide Web seminar Exam Prep for the performance and tuning.

  Thank you Roger

  
  

  If there was an an additional symbol, then maybe selftest software would also create reviews of preparation for the performance and tuning.

  Perhaps more people write books on their performance and tuning of certification.

  That hangs high hopes on a couple of letters.  The problem, however, is not perception, but the economy. Traditional publishers must sell a lot of books to make return on investment in the costs of creating a new. Until they beat X-thousand books, they are in the red. STS and transcend do not have publication costs, but they still have to pay someone (probably a very expensive one) to go through the mind-numbing task of creating a bunch of questions / answers that are enough like the real exam to be useful, but different enough to not be do not cheat. They should sell enough copies of the review of the practice of paying for that person, and then more overhead before making a profit.

  The main reason that I was able to create guides to study at the best prices is because they are printed on demand.  I don't have to recover the costs of printing a large number of copies in the hope they will sell. In addition, as I do all of the writing to editing to marketing of myself - it has no salaries to pay someone else. Despite this - I could easily do more initial money by running Oracle contract/work councils for the same number of hours that I put in my books and practice tests. I was called to joke with my wife I could do more by night shifts working underground, asking people if they want fries with that.

  The problem with an author or an editor of creating a book especially for an Oracle performance certification (say 1Z0-064 or 1Z0-117), is that there is so much work involved in this than the creation of a book of generic performance Oracle or Oracle SQL Tuning book, but the very act of this direct to the test severely limits the people who can buy it. Those who study for 1Z0 - 064 could well but a generic performance Oracle 12 c book.  However, people in the market for a book of performance 12 c but * not * intended to take certification 064 are not likely to buy this book even though she 1Z0 - 064 on the lid.

  There are not many hug of people with the knowledge and the ability to create these materials. This relatively small pool, it is there not much with the desire to do. This further reduced pool, there not much time for her - or at least of time they are willing to take more lucrative projects requiring their skills.

• two loop running concurrently, but double length

  Hi guys,.

  I try to read at the same time both ion Chamber. IC0 IC1 reading and reading have a loop of each.

  These two Subvi can run concurrently, but by the time the cost doubled again.

  Why?

  Thank you very much!

  

  Hi guys,.

  Thanks for your reply.

  I found the problem. I put the delay between write and read the com see the following figure. The timer was implemented with Resp.vi of IC101. After I moved out, my code can work well. It's the job

• In performance parameters, run FASTER, the OPTIONS of PERFORMANCE IN == ADVANCE = run FASTER, * it TOLD me TO DISABLE SOME SETTINGS, could you TELL ME WHAT SETTINGS I CAN UNCHECK.

  * Original title: TO RUN FASTER, IN PERFORMANCE OPTIONS

  in the properties of systems, performance parameters, run FASTER, the OPTIONS of PERFORMANCE IN == ADVANCE = run FASTER, * it TOLD me TO DISABLE SOME SETTINGS, could you TELL ME WHAT SETTINGS I CAN UNCHECK.

  When and where your computer is slow? Is the time required to start-up or after the reboot is complete? Browse the internet or work on files offline?

  Select Start, run, type msconfigand press the ENTER key or click on the Startup tab. You will see a window like the image below. What elements are there?

  In Windows 7, use Ctrl + Shift + Esc instead of Ctrl + Alt + Delete. It lets you in the Manager of tasks more quickly. Select Task Manager, performance, resource monitor tab and tab memory. What are the numbers for reserved equipment, in use, modification, sleep and free?

  Is your Windows 7 32-bit or 64-bit? The amount of RAM is installed?

  THE IMPACT OF THE COMPUTER ON THE PERFORMANCE SPECIFICATION
  http://www.gerryscomputertips.co.UK/performance1.htm

• Oracle performance management system

  Hi all

  Can someone help me on the implementation, the scope and assumptions about the Oracle performance management system?

  Hello

  If you want to learn the whole thing... go to this URL for R12.

  http://docs.Oracle.com/CD/B34956_01/current/HTML/docset.ht..findml

  Human resources, search for pl

  "Oracle HRMS effective Sourcing, deployment, and Talent Management Guide.

  with doc not. B31620-02.

  You can go through the talent management (which is up to 11i Performance management system).

  Is it to 11i... Click on the following link.

  http://docs.Oracle.com/CD/B25284_01/current/HTML/docset.html

  The study guide thoroughly and then ask specific questions.

  Kind regards
  Siva

• Intel SRCS16 on ESX4 - poor write performance

  Hello world!

  I need to change HD on one of my ESX hosts that have only 250 GB of total space, while we buy 5 HD 1 TB.

  This weekend, I put another control (Intel SRCS16) Raid and connect a 1 TB just to copy a VM of old HD to the new.

  Matrix Raid took less than 1 min.

  Creation of partition on the raid drive took less than 1 min. (with fdisk)

  Partition formatting took more than 3 hours (with mkfs.ext3)

  Test write performance with: time dd if = / dev/zero of=/mnt/share/test.test bs = 1 M count = 1024

  And took approximately 18 min. (1 Mb/s)

  Research on google, I find some useful links.

  On the Vmware communities, I found this: http://communities.VMware.com/thread/105552?start=0 & tstart = 0

  But it does not work.

  Someone has an idea?

  Thanks for your help

  SRCS16 Intel has some extremely poor write performance without cache enabled writable. For normal operations with write cache, you should have BBWC (battery backup). Enabling write without BBWC cache may lead to the loss of data in case of power failure.

  UPD. See http://www.intel.com/support/motherboards/server/srcs16/sb/cs-020502.htm.

• script to run several queries to find/replace

  Working on a big book of ID document (* .indb), I saved dozens of queries to search/replace (named "col01', 'col02' etc.) that I run on 'all documents '. For some reason, I have to run these queries several times a day.

  So I'm wondering if he has, or if someone out there will write: a script that runs these queries one after another. Given the size of the book, it may be necessary to provide the script to pause after each query (before it turns on with the next), but I'm just guessing.

  Any help is very appreciated!

  • Pre-installed FindChangeByList script (in the folder "Scripts Panel > Samples")
  • Script to change by queries to find
  • Multi-find/change script (commercial)

• Oracle Performance Tuning

  Hi all
  Can someone give me a good article explaining oracle performance, given gurd, rman and RAC.
  Although I worked on this subjects I am unable to answer a question in interview as
  How do you analyze your database? / How give you your DB... etc.
  
  First response would be appreciated.
  
  Thank you.
  
  
  Kind regards
  Rajini.

  Hi me;

  Please also check Learning Library for your documents need question
  http://Apex.Oracle.com/pls/Apex/f?p=44785:2:0:2:P2_GROUP_ID:1000

  Respect of
  HELIOS

• Oracle instance is running but sqlplus "virtue sysdba" says connected to idle.

  Hello
  Oracle RAC is running (started the CARS with "srvctl start database hacmp - d") but when I am trying to connect to the database using sqlplus 'virtue sysdba' he says 'connected to an idle instance.
  
  No idea what could be the problem here?
  
  I have exported ORACLE_SID.
  
  
  bash-3. $00 ps - ef | grep pmon
  Oracle 507946 798870 0 17:23:42 pts/0 0:00 grep pmon
  Oracle 872584 1 0 17:03:16 - 0:00 ora_pmon_hacmp1
  
  bash-3. $00 echo $ORACLE_HOME
  / u01/app/oracle /.
  bash-3. $ 00
  
  bash-3. $00 echo $ORACLE_SID
  hacmp1
  bash-3. $ 00
  
  bash-3. $ 00
  bash-3. $00 sqlplus "virtue sysdba".
  
  SQL * more: Release 10.2.0.4.0 - Production on Mon Sep 14 17:24:43 2009
  
  Copyright (c) 1982, 2007, Oracle. All rights reserved.
  
  Connect to an instance is idle.
  
  SQL

  Try to remove the ending ' / ' the ORACLE_HOME environment variable.

• Problem with Oracle performance

  Hi all
  
  I use Oracle database 11 g 2.
  
  I am facing a problem with one of my queries.
  
  
  Scenario is:
  I associate myself with 3 tables.
  The engine is saa_hist.saa_rep_product_activity (70 million lines).
  The second table is scmsa_hist.scmsa_sub_feature_trueup (900 million lines).
  The third table is saa_hist.saa_rep_subscriber_activity (35 million lines).
  
  The output of the query will be only about 120K.
  
  I'm majoriy of the columns in the first table. On the other hand, we get 2 columns and in table 3, we get a single column.
  
  When I join all 3 tables as it appears in the query runs for 6 hours. (The
  
  SELECT / * + PARALLEL_INDEX(R,SAA_REP_PRODUCT_ACTIVITY_BI,12)
  PARALLEL(SUB,24)
  PARALLEL_INDEX(RR,SAA_REP_SUBSCRIBER_ACTIVITY_I1,8)
  
  */
  R.*, sub.mrc, sub. RATEPLANCODE, rr.measure_group_id
  Saa_hist.saa_rep_product_activity r
  JOIN scmsa_hist.scmsa_sub_feature_trueup sub
  ON sub.job_log_id = r.job_log_id
  AND sub.service_number = r.msisdn
  AND sub.ban = r.account_number
  AND sub.soc_code = r.product_id_code
  LEFT OUTER JOIN saa_hist.saa_rep_subscriber_activity rr
  ON rr.quantity = 1
  AND rr.service_universal_id = r.service_universal_id
  AND rr.msisdn = r.msisdn
  AND rr.account_number = r.account_number
  WHERE r.billing_subsystem_id = 1
  AND 1 = 1
  AND r.batch_key BETWEEN 674 AND 675;
  
  The explain command plan looks good for this one and the cost is also okay(222K). I don't see any S-> data flow in the Section of IN-OUT of the explain plan P.
  
  -----------------------------------------------------------------------------------------------------------------
  | ID | Operation | Name | TQ | IN-OUT | PQ Distrib.
  -----------------------------------------------------------------------------------------------------------------
  | 0 | SELECT STATEMENT |
  | 1. COORDINATOR PX |
  | 2. PX SEND QC (RANDOM). : TQ10001 | Q1, 01 | P > S | QC (RAND) |
  | 3. NESTED EXTERNAL LOOPS | Q1, 01 | SVCP |
  | 4. HASH JOIN | Q1, 01 | SVCP |
  | 5. JOIN FILTER PART CREATE | : BF0000 | Q1, 01 | SVCP |
  | RECEIVE 8 2 PX | Q1, 01 | SVCP |
  | 7. PX SEND LOCAL BROADCAST | : TQ10000 | Q1 00 | P > P | BCST LOCAL |
  | 8. PX HASH PARTITION ALL | Q1 00 | ISSUE |
  | 9. TABLE ACCESS BY LOCAL INDEX ROWID | SAA_REP_PRODUCT_ACTIVITY | Q1 00 | SVCP |
  | 10. CONVERSION OF BITMAP IN ROWID | Q1 00 | SVCP |
  | 11. BITMAP INDEX RANGE SCAN | SAA_REP_PRODUCT_ACTIVITY_BI | Q1 00 | SVCP |
  | 12. FILTER-HASH PARTITION PX JOIN | Q1, 01 | ISSUE |
  | 13. TABLE ACCESS FULL | SCMSA_SUB_FEATURE_TRUEUP | Q1, 01 | SVCP |
  | 14. RANGE OF PARTITION ALL THE | Q1, 01 | SVCP |
  | 15. HASH PARTITION ALL | Q1, 01 | SVCP |
  | 16. TABLE ACCESS BY LOCAL INDEX ROWID | SAA_REP_SUBSCRIBER_ACTIVITY | Q1, 01 | SVCP |
  | 17. INDEX RANGE SCAN | SAA_REP_SUBSCRIBER_ACTIVITY_I1 | Q1, 01 | SVCP |
  -----------------------------------------------------------------------------------------------------------------
  
  But when I joined the first two tables and run as shown below, the request is get executed in 13 Minutes.
  
  SELECT / * + PARALLEL_INDEX(R,SAA_REP_PRODUCT_ACTIVITY_BI,12)
  PARALLEL(SUB,24)
  */
  R.*, sub.mrc, sub. RATEPLANCODE, rr.measure_group_id
  Saa_hist.saa_rep_product_activity r
  JOIN scmsa_hist.scmsa_sub_feature_trueup sub
  ON sub.job_log_id = r.job_log_id
  AND sub.service_number = r.msisdn
  AND sub.ban = r.account_number
  AND sub.soc_code = r.product_id_code;
  
  Therefore, I created a temporary table with this data and made a left outer join with the other 3 table. It runs a few seconds and I can say all the data I am able to take less than 15 minutes to 6 hours.
  
  So my question is that why Oracle is a strange behavior.
  
  I tried all approaches to avoid this kind of behavior methods but unable to produce the same thing as how I did in the second approach.
  
  For example, I joined the first two tables and use NO_MERGE hint with the third table and always NO_MERGE, PUSH_PRED's not going anywhere. I tried Inline view and it doesn't give me any improvement in performance.
  I'm tempted to do it using WITH Clause with two tables and then joined the data set with the third table. But still does not.
  I tried hint ORDINATE and yet so big difference.
  I even tried to force some clues which are presentm, but the 50Million cost ranges from 200K.
  I think I'm exhausted all the possibilities.
  
  I'd appreciate any Oracle guru can help me what is the issue.
  
  Thank you
  NKM

  You could start trying to impose (with extensive predicates) what you discover or believe work better.
  In this case I usually found the ansi syntax not useful and often the optimizer ignore my intentions:(mais il n'y a pas de mal essayer)

  select x.*,
         rr.measure_group_id
    from (select r.*
                 sub.mrc,
                 sub.rateplancode,
            from (select *
                    from saa_hist.saa_rep_product_activity
                   where billing_subsystem_id = 1
                     and batch_key between 674 and 675
                 ) r,
                 scmsa_hist.scmsa_sub_feature_trueup sub
           where sub.job_log_id = r.job_log_id
             and sub.service_number = r.msisdn
             and sub.ban = r.account_number
             and sub.soc_code = r.product_id_code
         ) x,
         (select service_universal_id,
                 account_number,
                 msisdn,
                 measure_group_id
            from saa_hist.saa_rep_subscriber_activity
           where quantity = 1
         ) rr
   where x.service_universal_id = rr.service_universal_id(+)
     and x.msisdn = rr.msisdn(+)
     and x.account_number = rr.account_number(+)
  

  Concerning

  Etbin

• Poor search performance clustered with table of singles (100 k)

  In a data entry project, I created a data type to store regular data. I have a cluster (see photo) which consists of a name, StringData, Numericaldata and ArrayOfSinglesData (predefined 100points k). Thus, every piece of data can be a scalar string, scalar number or an array of numbers.

  I then do a picture of this for all of my variables, maybe 150 or more.

  Works well, but the performance when adding new data (search for the exact name) is pretty poor.  Very high CPU and many even page faults for the slow connection like 1 to 10 times per second.

  The entrance to the data add method is an array of (stringname, DataString). So for each new data I loop over all the variables, find the one with the correct name string and add new data (replce the scalar chain and the number, add to arrayOfSingles).

  Given that I have 150 variable, as I thought a linear search should be ok, but probably the program moves also around 100 k singles for each comparison, or it's my idea at least.

  Can you suggest which is a good thing to do to improve performance?

  I added a search binary smarter, but it was necessary to sort variables, and since I have two names for each variable (called internal and external) that I use when you add, it helped only for half of the research.

  If I find the time I would like to divide the cluster so that all ArrayOfSingles are separated, but maybe you can tell me immediately if it is useful?

  (Labview, 2012sp1)

  Like Gerds link watch I suppose you're looping in the table looking for your name?

  I just tested, and using DVR to extract only the ID and grabbing then a single element are fast enough. I will attach my test so you can try it for yourself.

  /Y

• Running parallel queries in 11 GR 2

  I'm a little confused about parallel execution in GR 11, 2.

  Lets say I have the Setup below.

  parallel_servers_target = 4;

  PARALLEL_MAX_SERVERS = 16;

  parallel_degree_limit = CPU

  parallel_threads_per_cpu = 2

  parallel_adaptive_multi_user = TRUE

  N ° of UC = 16 CPU, 2 cores each.

  The Test table degree = DEFAULT

  Example query:

  Select * from Test by 1,2,3;

  Lets say I want to run 5 in parallel sql statements that require 4 parallel workers each.

  PARALLEL_DEGREE_POLICY = AUTO:

  What of her below happen?

  (a) a statement was going to be executed and the rest four would be queued because the first statement itself uses 4 parllel workers what amounts to our parameter of parallel_servers_target.

  or

  (b) four statements would be carried out using 4 workers each and the fifth would be queued because it exceeds the parallel_max_servers setting if it were to be executed.

  or

  Please explain what could happen if not a) or b).

  PARALLEL_DEGREE_POLICY = MANUAL:

  What of her below happen?

  (a) four statements would be managed (using 4 parallel slaves each) and the fifth will be held in the series.

  or

  (b) all five would be using parallel workers. If so, please explain how the No. parallel to each statement slaves were to be distributed.

  or

  (c) it would first try to allocate 16 sons in the first statement. Later it will adjust accordingly and reduce parallel workers allocated to each statement and end up running two 4 statements with parallel workers or all the five workers parallel.

  I appreciate really all of your answers on this.

  UC = 60 (perhaps CPU_COUNT = 60), the number of cores = 15 - so you have four hardware threads by heart.

  The DOP default value is calculated as PARALLEL_THREADS_PER_CPU * CPU_COUNT (in the single instance databases) - in your case, the DOP default is 2 * 60 = 120.

  See this page 20 (http://www.oracle.com/technetwork/database/bi-datawarehousing/twp-parallel-execution-fundamentals-133639.pdf) :

  Manually set the degree of parallelism

  With PARALLEL_DEGREE_POLICY set to MANUAL Auto DOP feature is disabled and the end user must manage the use of parallel execution in the system.

  You can either apply the default value of the so-called DOP or a specific fixed value for the declaration of principles on a session, statement or object level.

  LACK of parallelism

  ...

  Fixed degree of parallelism (DOP)

  Unlike the LACK of parallelism, a specific DOP can be asked to the Oracle database. You can use one of the following ways to get a fixed DOP.

  1. define a fixed DOP for objects. ALTER TABLE 8 PARALLELS customers. ALTER TABLE 16 PARALLELS sales;

  2. use the suspicion level parallel (integer) statement. SELECT / * + parallel (8) * / COUNT (*) OF customers.

  3. use the level indicator parallel (table_name, whole) object. SELECT / * + parallel (8 customers) * / COUNT (*) OF customers.

  Note that the parameters in the table on the #1 above give you only the fixed DOP in manual mode and mode limited, namely when PARALLEL_DEGREE_POLICY is set to MANUAL or LIMITED.

  All table decoration will be ignored in mode (AUTO or ADAPTIVE) AutoDOP.

  Also note that for example # 1 parameters, Oracle will choose the DOP requested as follows:

  -The queries accessing the CUSTOMERS table use a DOP asked 8.

  -Accessing the SALES table queries ask a DOP of 16.

  -Queries accessing both the SALES and the CUSTOMERS table will be processed with a DOP of 16. Oracle uses the upper DOP2.

  The number of allocated PX servers can become always twice the requested DOP in the case of parallel processing requires two sets of PX server for treatment of producer/consumer.

  And see on page 25:

  Parameter: PARALLEL_MAX_SERVERS

  Default value: PARALLEL_THREADS_PER_CPU * CPU_COUNT * concurrent_parallel_users * 5

  Suggested value: default

  Explanations: Sets the maximum number of servers in parallel execution that can allocate an instance of database. This is a hard limit and cannot be exceeded.

  So, put PARALLEL_MAX_SERVERS > 120.

  In fact, it would be better to change the PARALLEL_THREADS_PER_CPU = 1 parameter (default is 2). See this page 25:

  Parameter: PARALLEL_THREADS_ PER_CPU

  Default value: 2

  Suggested value: 1 for platforms with hyperthreading enabled, 2 for other platforms

  Explanations: Number of parallel processes that a processor can handle in parallel execution

  Then the default DOP would be PARALLEL_THREADS_PER_CPU * CPU_COUNT = 1 * 60 = 60.

  In this case, set PARALLEL_MAX_SERVERS > 60.

  Kind regards

  Zlatko

• Causes of white space, tab, and line breaks slow in running the queries? What is the main reason behind it?

  Hi all

  
  

  I am facing a weird problem in the SQL query. Taken, I wrote the request with the spaces, tabs and line breaks after execution it slows down the database and finally, I have to restart the database again. But after removing the spaces the application went very well. I don't know what the problem causing this slow run, but after the removal of whitespace to query everything works fine. I've also confirmed plans to explain the two after the removal of whitespace and previously had the same results.

  I use the TOOL of DEVLOPER PL/SQL, SQL developer, Toad and Oracle 11 g.

  ***************************************************************

  For example. (in reference to my original query):

  Select

  *

  Of

  Double;

  ****************************************************************

  (After removing whitespace)

  Select * twice;

  *****************************************************************

  What would be the reason for white spaces causing slow performance?

  Please let me know if you need additional information on my question.

  
  

  Thanks in advance,

  HP

  If you make the first request without spaces and the second by spaces you will see, the second is even faster...

  The reason why are cached revenge...

  HTH

• Oracle OpenScript cannot run more than 3 Scripts in a session

  Dear,

  After you create the function of Web test scripts and fusion in a Script to help run the Script tool, Oracle openscript run only 2 or three scripts and display of the browser the message below:

  OpenScript: slowed the mouse position:

  and does not perform the remaining scripts

  The details of my Machine are:

  OpenScript Version: 12.2.0.1 Build 288

  Windows 7 32 bit

  Processor I5

  2 GB of RAM

  Internet explore 8

  Waiting for your comments

  work after the replacement of the browser.close () to close a session with the web.window("path").close option.