Query result cache

Similar Questions

• The query results cached?

  Hi guys,.
  
  I have this page that calls the same query (on a database link!) four times! First when an initial LOV is generated (and the page loads), again for an another LOV based on the original and finally for the generated report based on the 2 selections. It's insanely slow (30-40 seconds on each load ~) and also: when you try to sort a report interactive by clicking on the column name, and then queries the database again (I assumed that this part has been cached but apparently not). I so need to caching the results of an initial request and then just limit the results locally. How can I do this? Guides? Examples?
  
  Thanks for help.
  
  Mike

  I tried this example based on a view:

  CREATE MATERIALIZED VIEW MV_TEST2
       REFRESH COMPLETE
       START WITH SYSDATE
       NEXT  SYSDATE + 1/48
       WITH ROWID
       AS SELECT * FROM test1;
  

  DISCOUNT COMPLETE - The complete refresh recreates together materialized view.
  START WITH SYSDATE - run now
  NEXT SYSDATE + 1/48 - run again in half an hour
  WITH ROWID - I think that this option is important if you are using a partial refresh of the view.
  AS SELECT * FROM test1; -test1 is a point of view:

  CREATE OR REPLACE VIEW TEST1 AS
  SELECT st_id, st_name
      FROM aaw_solution_tree;
  

  Index of the column are still possible? I do not know:
  Indexing:! in regards to the MV about 10 gr 2 Jonathan Lewis wrote. ... you are allowed to create indexes on tables that sit under materialized views - just don't make them unique indexes

  How much freedom is there to set the refresh rate?
  What kind of discount do you need?

  Another useful link: [http://asktom.oracle.com/pls/ask/search?p_string=materialized+view | http://asktom.oracle.com/pls/ask/search?p_string=materialized+view]

  It will be useful.

  Tobias

• Query results cached in DBlink (of 8.1.7.4.1-> 9.2.0.8)

  I met a strange problem today...
  
  
  I have 2 versions of oracle version 8.1.7.4.1 running on Win2k and a 9.2.0.7 running on a win2k3 server... Both versions are standard edition 32 bit OS
  
  version 9i has been improved from 9.2.0.6 to 9.2.0.8 and begun the issued after that.
  
  1 link to DB on a database of 8 with a link to Oracle 9
  
  Query:
  

  sql> select family_name from hrm_employee@db92 where <columnname>='1234' and <columnname>='MARAN'
  
  This result is one row
  
  but when i try again with a different values to the column it still throws me the same result set which is supposed to return no rows
  
  sql> select family_name from hrm_employee@db92 where <columnname>='1234' and <columnname>='MARAN DUMMYVALUE'

  Returns the same line... that was returned by the previous query. but it works perfectly when I connect from an another 9.2.07 against the 9.2.0.8 db
  
  I raised to support oracle but it looks like a bug?
  
  If someone met before...
  
  Published by: Maran Viswarayar January 12, 2010 11:35
  
  Published by: Maran Viswarayar January 12, 2010 11:42

  Yes, I remember some weird side effect when the value of this parameter is not CORRECT on Oracle<>

  Nicolas.

• Using the ODP client result cache still shows it as running on AWR, is the work of cache?

  Hello

  I use the Oracle 12 manged customer ODP.net connected to a database of GR 11, 2. I recently went on the cache of results with great effectiveness.

  So, I thought I would try the result to the client cache. I have the setting turned on and restarted my db, but my questions always appear in the graphs of Enterprise manager and the AWR report. I would have thought if the query has been cached on the client, that it does not show.

  How will I know if it works or not?

  Thanks in advance.

  Adrian.

  Re-reading the original question, I missed the fact that you use managed ODP.NET. The client result cache does not support managed ODP.NET. Only unmanaged ODP.NET is because it uses the capacity, within the OIC, that uses the unmanaged code.

  I apologize for the misunderstanding.

• Update statement can be used depending on the result cache?

  Hi all

  I stumbled on an interesting question. UPDATE statement is used in the following function, even if the feature is created RESULT_CACHE. It seems illogical. I was wondering, is it possible?

  If so, why RESULT_CACHE is used?  Because it means "do not execute the function, look for the result of the hash table and return the results to the user ' if the UPDATE statement, so doesn't mean it should run for each call and use with RESULT_CACHE and UPDATE/DELETE/MERGE is illogical or wrong?

  Thanks for your help.

  FUNCTION to CREATE or REPLACE plch_get_data (FULL id_in)

  RETURN VARCHAR2

  RESULT_CACHE

  IS

  BEGIN

  Dbms_output.put_line ('run');

  UPDATE plch_data

  SET nm = UPPER (nm)

  WHERE id = id_in;

  COMMIT;

  RETURN "SUPERIOR."

  END;

  /

  Yes, it's the challenge of PL/SQL quiz - and the point of the quiz is only automatic relies on parsing by Oracle only supports arrays of account that is QUESTIONED.

  The tables that are affected by DML not a query as an update do not figure in automatic invalidation of a cache.

  I was hoping that the explanations given in the questionnaire itself would not lead to a thread - rather it would provide a pleasant and clear answer. So feel free to let me know if you think otherwise.

  As to why Oracle would let you include DML not a query inside a function of caching of result, well... I can easily accept that you shouldn't do this - in general. But I don't see that we should make it impossible. It would probably be a good candidate for an another PLW (PL/SQL warning), as in:

  "Non-requetes DML in a result caching function will not affect caching and can result in unexpected results."

• Using the Client result Cache

  Hi all

  I have a question about the use of the result of the client cache but first of all, here is the information on my database:

  SQL> SELECT * FROM v$version;
  
  BANNER
  --------------------------------------------------------------------------------
  Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
  PL/SQL Release 11.2.0.1.0 - Production
  CORE    11.2.0.1.0    Production
  TNS for Linux: Version 11.2.0.1.0 - Production
  NLSRTL Version 11.2.0.1.0 - Production
  
  

  I installed the Oracle Client Tools on my clientmachine. There a tnsnames.ora like this:

  ORCL =
    (DESCRIPTION =
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = TCP)(HOST = FEA11-119SRV)(PORT = 1521))
      )
      (CONNECT_DATA =
        (SERVICE_NAME = orcl.local)
      )
    )
  
  

  The result of the Client Cache is configured in the sqlnet.ora.

  ADR_BASE = /u01/app/oracle
  OCI_RESULT_CACHE_MAX_SIZE= 2048000
  OCI_RESULT_CACHE_MAX_RSET_SIZE = 1024000
  NAMES.DIRECTORY_PATH = (EZCONNECT, TNSNAMES, HOSTNAME)
  
  

  The Databaseserver is configured like this:

  
  SQL> col name format a30
  SQL> col value format a30
  SQL> SELECT name, value
    2  FROM   v$system_parameter
    3  WHERE  name LIKE '%result_cache%';
  
  NAME                   VALUE
  ------------------------------ ------------------------------
  result_cache_mode           MANUAL
  result_cache_max_size           0
  result_cache_max_result        5
  result_cache_remote_expiration 0
  client_result_cache_size       1073741824
  client_result_cache_lag        3000
  
  

  After you set up a client's server, I try the SQL query on the clientmachine followign:

  sqlplus oracle/password@orcl
  
  set autotrace traceonly explain
  
  SELECT /*+ result_cache */ *
  FROM testtable;
  

  You cannot enable the customer with SQL result cache * more, no matter what you do with parameters. If you install the CD examples, you will find examples of code in cdemoqc.c and cdemoqc2.c that shows how with the OCI calls.

  You probably want to use the result of the server cache, there is a full demo of this here,

  Using the SQL Oracle 11g result Cache

  --

  John Watson

  Oracle Certified Master s/n

• Result Cache in oracle 11g release 2.0.2.0

  All,
  
  A few months ago I tried function result Cache 11 g version 2.0.1.0. The oracle of time mentioned that they have 3 ways to use this feature by setting the parameter result_cache_mode AUTOMATIC, MANUAL or FORCE. With AUTO (result_cache_mode = AUTO) mode you can identify the queries that can benefit from the cache of the result. I would like to know what views or function I see where he captures this information.
  In 11.2.0.1.0 documentation he was making reference to certain points of view (totally forgot the names of views) and said: those that will be available in 11.2.0.2.0.
  
  Did anyone know about this?
  
  Thanks in advance.

  user2486753 wrote:
  Although doc oracle didn't tell on AUTO it is still accepts this value and caches the result of the SQL query if you add / * + cache result * / allusion to the SQL.
  I know that they mentioned in 112010 somewhere that will help users find what SQL is good candidates for caching by running the app for a while. I couldn't just fine there.

  If we add AUTO and always passing the tip, this is the same behavior as MANUAL. I'm not sure I've seen the value of anywhere and neither its there in the docs too.

  HTH
  Aman...

• Keep the query results set all paging

  I have a query back thouands of records. For do not impact performance display and paging, which would be the best way without use of query caching? Is it possible to keep the first recordset query time she and her use in Exchange?

  Load balancing definitely throws a wrinkle by using shared memory.  There are ways to address these problems, but they all come with the pros and cons, just basically choose what pros like you and what the disadvantages that you can live with.

  I.E. you can use sessions.  So when a user first access your application, all future queries are sent to the same server that processes requests for first.  But this means that if this server will see, all users currently glued to it are s.o.l.

  I saw other people using databases for the persistent memory.  But then, you have the LAG involved in obtaining data from the database on every request, although it should be easy fast queries to do this.  Side pro, it does not matter which server processes the request.

  It would be different then just re - run the original query every time, but rather store the original query results into a temporary table space so that any complexity which may exist should not be repeated.

• TimesTen and Oracle 11 g result cache

  Oracle 11g introduces the concept of cache the result whereby the result set of frequently executed queries are stored in the cache and used later when other users ask the same query. This is different from the caching of the data blocks and exceuting the query over and over again.
  
  Tom Kyte calls this materialized view just-in-time, in which case the results are dynamically evaluated without DBA intervention
  
  http://www.Oracle.com/technology/oramag/Oracle/07-Sep/o57asktom.html
  
  My point is that, given utilities like result_cache and possibility to use Solid State Disks in Oracle to accelerate physics I/O etc. is necessary for a product like TimesTen? It seems to me he could asdd just another layer of complexity?

  Oracle cache there result a useful tool, but it is significantly different from TimesTen. My understanding of the Oracle result cache is caching the results for rarely changing data such as search (currencies/identification code) tables, data reference that do not change often (list of counterparties) etc. It would be pointless to cache the result set when the data changes frequently.

  There is also another argument for the cache of SQL result that if you hit high on your CPU usage and that you have enough memory, then you can cache some defined thus results save on your cycles CPU.

  In view of the arguments on the wired RDBMS and drives (SSD), we can talk all day, but with SSDS does not eliminate the consideration for physical i/o optimizer. A table scan is a table scan if the data resides on disk SCSI or SSD. SSD will be faster, but we are still running physical IOs.

  What about TimesTen, the positioning of the product is different. TimesTen is closer to average than Oracle. It is designed to work closely at the application layer, while Oracle aims much more wide. There is no way one can replace TimesTen with any RDBMS wired for volumes of moderate and real-time response. Demand for the result cache has been around for a while. In areas like trading program and data on the market when the underlying data changes quickly, TimesTen will come handy, because time real/provisional data and calculations must be made almost in real time, with fewer complications of the execution engine. I don't see how we can deploy the result cache in this scenario. Due to underlying data change, Oracle is obliged to calculate queries almost everytime and the cache of result will be just wasted.

  Hope this helps,

  Mich

• Components catalog SQL query and caching

  Hello
  
  It seems that SQL query result sets components catalog caching and go to the DB only after restarting the engine. Is it possible to disable this caching and force SQL queries to run each time?
  
  Thank you.
  Nick.

  Hi Nick,

  Take a look at the log file of your engine. To execute SQL queries when running logic. See if you get any 'error', 'Serious' or 'Warning' engine log messages when you run your query.

  Hope this helps,
  Dan

• Need help to build the query/pl-sql block to get the query result and the name of column from DB table in the form of key-value pairs.

  Hi Experts,

  I have a DB table has columns of more than 50.

  I question this table, it should only return one line at any time. as sqldeveloper below image.

  

  here, I need to build block pl/sql-query, Discover the column in the table as a key and query result as values.

  Eg:     Key                         -  Value

  TASK_EVENT_ID - 1765

  EVENT_TYPE - ASR_UPDATE

  ... etc until all of the columns in my table.

  Experts please comment on that point, appreciate your help on this.

  Thank you

  -Vincent.

  Here is an approach using DBMS_SQL to iterate over the columns of key / value to assign... (Little code snipped for brevity)

  create or replace procedure (task_expired)

  v_store_id in full,

  v_task_action_id in full,

  v_job_id in full

  )

  as

  -[SNIP code...]

  v_sql VARCHAR2 (4000): = ' select * from my_table where PK = 123';  -Your SQL here!

  v_v_val VARCHAR2 (4000);

  v_n_val NUMBER;

  v_d_val DATE;

  v_ret NUMBER;

  c NUMBER;

  d NUMBER;

  col_cnt INTEGER.

  f BOOLEAN;

  rec_tab DBMS_SQL. DESC_TAB;

  col_num NUMBER;

  vAsString VARCHAR2 (4000);

  BEGIN

  -[SNIP code...]

  Message_properties. CORRELATION: = "EDF_EVENT";

  MSG: = SYS. AQ$ _JMS_BYTES_MESSAGE. Construct();

  Msg.set_string_property ('queueName', ' shipping/csi_cth');

  Msg.set_string_property ('MODE', 'CR8');

  c: = DBMS_SQL. OPEN_CURSOR;

  DBMS_SQL. PARSE (c, v_sql, DBMS_SQL. NATIVE);

  d: = DBMS_SQL. Execute (c);

  DBMS_SQL. DESCRIBE_COLUMNS (c, col_cnt, rec_tab);

  1.col_cnt J

  LOOP

  CASE rec_tab (j) .col_type

  WHEN 2 THEN

  DBMS_SQL. DEFINE_COLUMN (c, j, v_n_val);      -Number

  WHEN 12 CAN

  DBMS_SQL. DEFINE_COLUMN (c, j, v_d_val);      -Date

  ON THE OTHER

  DBMS_SQL. DEFINE_COLUMN (c, j, v_v_val, 2000);   -Else treat as varchar2

  END CASE;

  END LOOP;

  LOOP

  v_ret: = DBMS_SQL. FETCH_ROWS (c);

  WHEN OUTPUT v_ret = 0;

  1.col_cnt J

  LOOP

  -Fetch each column to the correct data type based on coltype

  CASE rec_tab (j) .col_type

  WHEN 2 THEN

  DBMS_SQL. COLUMN_VALUE (c, j, v_n_val);

  vAsString: = to_char (v_n_val);

  WHEN 12 CAN

  DBMS_SQL. COLUMN_VALUE (c, j, v_d_val);

  vAsString: = to_char (v_d_val, ' DD/MM/YYYY HH24:MI:SS');

  ON THE OTHER

  DBMS_SQL. COLUMN_VALUE (c, j, v_v_val);

  vAsString: = v_v_val;

  END CASE;

  Msg.set_string_property (rec_tab (j) .col_name, vAsString);

  END LOOP;

  END LOOP;

  DBMS_SQL. CLOSE_CURSOR (c);

  DBMS_AQ. ENQUEUE (queue_name-online 'cbus.aqjms_common',

  Enqueue_options => Enqueue_options,

  Message_properties => Message_properties,

  Payload-online msg,

  Msgid => Message_handle);

  dbms_output.put_line ('00 Msgid =' |) Message_handle);

  dbms_output.put_line('===Done=');

  -[SNIP code...]

  END;

  /

• How to get SQL Developer 4.1 to display the query result?

  SQL DEV 4.0.3.16.84 execution of a SELECT statement is stored in a .sql file causes the window of the query result of pop-up indicating the result of a SELECT statement.  SQL DEV 4.1 is not to do so.

  Oops on my part; my SQL formatting has been the root cause.  I got used to insert split without any character of comment lines in front of them.  For example:

  ==================================================

  Select funny business;

  The foregoing indicates the behavior described.

  --==================================================

  Select funny business;

  With the dash double comment shows, the SQL works very well.

  This is different than previous versions of SQL Developer, but is probably better functionality.  I'll try to mark it as answered, but anyone with the admin should certainly do so if I can't.

• Need help to understand the query result

  Hi gurus

  I was reading one of the question here in this forum and its link is below:

  Query required for scenario

  I had some confusion related to this code and don't understand the logic of the out put, see query below:

  Query

  with sub_services as

  (

  Select su_seq 12323, 'HLR1' so_id, 1 seq Union double all the

  Select su_seq 12323, "HLR2' so_id, seq 2 Union double all the

  Select su_seq 12323, "A09" so_id, seq 3 of all the double union

  Select su_seq 12333, "MO1" so_id, seq 4 Union double all the

  Select su_seq 12333, "MO2' so_id, seq 5 Union double all the

  Select su_seq 12333, "A09" so_id, 6 seq in union double all the

  Select su_seq 12333, 'M0CR' so_id, seq 7 Union double all the

  Select su_seq 12999, "LOL1' so_id, seq 8 Union double all the

  Select su_seq 12999, "LOL2' so_id, seq 9 double

  )

  Select *.

  of sub_services b

  where exists (select 1 from sub_services

  where su_seq = b.su_seq

  and so_id = 'A09.

  )

  order by 2;

  The query result

  12323 A09 3

  12333 6 A09

  12323 HLR1 1

  12323 HLR2 2

  12333 M0CR 7

  12333 4 MO1

  12333 5 MO2

  According to my understanding, the above query should return records in red only because of her is below command

  It exists (select 1 from sub_services

  where su_seq = b.su_seq

  and so_id = 'A09.

  but don't know why he's back 7 files, can someone help me understand the result...

  It is query is functionally identical to the PL/SQL block, but much more effective.

  declare

  number of l_res;

  Start

  for line (select *)

  sub_services) loop

  Start

  Select 1 from l_res

  of sub_services

  where su_seq = row.su_seq and

  so_id = "A09" and

  rownum = 1;

  exception when

  NO_DATA_FOUND then

  null;

  end;

  end loop;

  end;

  Essentially every row in the outer query are tested against him exists query.  Given the correlation between two requests is based only on su_seq each line with a su_seq value returned by him is returned in the output.

  Another way to think he uses instead a join condition.  This query is equivalent to the query to exist

  Select the main

  of main sub_services

  Join select (separate su_seq

  of sub_services

  where so_id = "A09") cond

  We main.su_seq = cond.su_seq;

  John

• Oracle result Cache problem

  Hello

  In our database, we have the following settings for caching the result:

  result_cache_max_result integer 5

  result_cache_max_size great whole 256256 K

  result_cache_mode chain MANUAL

  result_cache_remote_expiration integer 0

  But we use indicators no results hidden in questions, or tables that are enabled for result caching.

  So from my understanding, caching the result should not happen.

  But we met a performance problem, and when I checked the AWR report I found this:

  

  And also I found this activity latch:

  

  However, just set result_cache_max_size = 0 fixed to all our questions. Wait no more. Performance as expected.

  But how is that possible? I thought with result_cache_mode = manual and not using indicators of result cache or table modes,.

  There should be no result at all caching? Any max_size choose?

  Can someone help me understand what is happening here?

  Thank you and greet.

  Daniel

  tylerv wrote:

  The original poster was not to the point and they actually understand the docs.

  Sorry - I'm not okay. As I went to the docs are very EXPLICIT about how to disable the cache of result. If you turn it off, then it is enabled.

  The fact that they have tried to do is that when result_cache_mode is set to the default value of the MANUAL, because they showed, it is, while according to the docs 12 c only the results of queries that use the RESULT_CACHE indicator must be stored in the cache of result

  Maybe that's where you and OP have misunderstanding. on the RESULT_CACHE_MODE parameter and the parameter values. This seems to be a common misunderstanding.perhaps, because people are not read ALL the relevant articles of the RFSO and only looking at the section for this parameter.

  There is NOTHING in the doc of 12 c for this parameter that shows ANY correlation to determine whether the cache is enabled or not.

  Refer to the section "Setting the Mode of the result Cache"

  https://docs.Oracle.com/database/121/TGDBA/tune_result_cache.htm#CDEJCGJD

  As shown in table 15-3 there are ONLY two settings: MANUAL and FORCE. No adjustment said ANYTHING on toggle or if the cache is ENABLED or DISABLED.

  ... and they were not aware of all the queries that they had used this trick on.  When you're ready to values by default and don't use don't not result_cache indicator (as much), it's odd to find the cache of your result are so full.

  OK - the USER does not use the indicator result-cache to cache the results of the queries USERS. But if the cache is enabled (size not equal to zero) that according to the quote from the doc above has NO effect on the use of Oracle to cache for queries or non-deterministic function calls.

  I think a lot of people may be jumping on the server I mentioned first that article explicitly states that you must set the size to zero to disable hide and seek only to the section on setting the parameters.

  And that the "setting mode" section is confusing as it includes a paragraph that uses the expression "If you choose to activate the result cache.

  That subsection and the r-phrase likely cause people to think that the setting and the off/on State of the cache are one and the same.

  When the result cache is enabled, the database also highlights cache queries that call non-deterministic functions of PL/SQL. When caching SELECT statements that call functions, result cache dependencies of data for PL/SQL functions and titles of database objects. However, if the function uses the data that are not followed (such as sequences, SYSDATE , SYS_CONTEXT and package variables), uses the cache of result on queries that call this function can produce stale results. In this regard, the behavior of the cache of result is identical to the caching of PL/SQL functions. So always consider the accuracy of the data, as well as performance, if you choose to activate the result cache.

  But if all associated sections are interpreted in their context, it should clear up the misunderstanding.

  If you have a support contract, you can submit a request to have the documentation changed to update mode setting (table 15-3) article to explicitly declare the parameter mode does NOT determine if the cache is enabled or not.

  .

• Result Cache not used

  Hello

  I'm on Oracle Database 11 g Release 11.2.0.3.0 and Application Express 4.2.2.00.11

  How to use trick cache result?

  How do you get the same result shown in Figuree 3 (apex.oracle.com Application Express 4.2.6.00.03) ?

  I try to use the result cache but explain plan does not appear to be used (Figure 1)

  Then I tried ' alter table STUDENT result_cache (FORCE mode);  In figure 2, you could see that column in the table of students result_cache was changed to 'force' but explain the plan shows yet cache result being used.

  Figure 1
  

  Figure 2

  

  Figure 3 (from apex.oracle.com workspace)

  

  Hello

  Thank you all for your answers.

  According to Oracle Help Center (11 g release 2) this is a feature reserved for Enterprise Edition. And Apex.oracle.com from today using Oracle Database 11 g Enterprise Edition Release 11.2.0.4.0 - 64 bit Production.

  While I'm using Oracle Database 11 g Release 11.2.0.3.0 - 64 bit Production it is why in the screenshots above I could not duplicate the result in apex.oracle.com

  https://docs.Oracle.com/CD/E11882_01/license.112/e47877/editions.htm#DBLIC116