Performance with unspecific where clause

Hi gurus,
at the moment I do have a sql statement on a view with an unspecific where clause like
select from <view> where <textfield> like %whatIsearch%
The text field is not in the key fields which are used to create the view. An index on <textfield> does not help, because the where clause starts with a %.
Other databases like Oracle do finish the statement within seconds, maxdb needs minutes.
I there a possibility to speedup the statement in MaxDB? (Besides telling the user to use better qualified statments.)
Why is Oracle that fast?
Thanks for you help.
Best regards
Christian G

> Other databases like Oracle do finish the statement within seconds, maxdb needs minutes.
> I there a possibility to speedup the statement in MaxDB? (Besides telling the user to use better qualified statments.)
> Why is Oracle that fast?
Hi Christian,
In that case Oracle can take advantage from beeing able to brute-force read many blocks at once (aka multiblock read).
When we assume that there is an index on the oracle database available, then Oracle will likely decide to read all blocks of that index into the cache and look for matches then. It's called Fast Full Scan.
It's not a very efficient method to address specific rows, but for this requirement it works well.
Anyhow, you should be aware that this way of evaluating rows does not scale very well - in fact it get's more expensive with every block the index grows.
MaxDB cannot easily read all blocks in a row, due to the fact that the pages are mixed over all data volumes. This way of storing data eliminates the need for reorganisations and evens out I/O traffic, but it comes for the price of beeing less performant when people use such unefficient predicates.
Because of this and the way who indexes work in MaxDB (primary keys instead of rowids), MaxDB does only consider an index access for like conditions that start with a '%' when the query can be answered by only accessing the index (index only access).
What you may try out to improve the situation is to activate the experimental read-ahead or prefetch feature, which is currently available in MaxDB 7.6. only (not in 7.5 or in 7.7).
By setting READAHEAD_TABLE_THRESHOLD to a value >0, say 128, MaxDB can choose to
perform table scans (no index scans!) in parallel with multiple server tasks for all table scans that are expected to pass the threshold (unit here is pages as visible in the execution plan of your statement).
That way the usertask running the query can work on checking the data in the pages, while the server task load the pages into the cache.
Another approach would be to have the DB Cache big enough so that most of the table would be found in the cache.
regards,
Lars

Similar Messages

How to create a procedure to output REF CURSOR with any WHERE clause?

I have an requirement like this: I have huge query which need to reuse in my code more than 10 times. This SQL has about 50 lines. Thing is for those 10 odd times sometimes the WHERE clause changes (columns are the same). So I cannot create a view since SQL is not static.
I thought of writing a procedure with a WHERE_CLAUSE input para. I output a sys refcursor by adding the where clause. But I can't do it since you cannot add a where clause like that.
i.e.
PROCEDURE dynamyic_query (p_where_clause IN VARCHAR2, p_out_query OUT SYS_REFCURSOR ) IS
BEGIN
OPEN p_out_query FOR SELECT ......... FROM table WHERE || ' ' || p_where_clause;
END;The above gives error.
How to handle a situation like this???? Any help would be greatly appreciated.

I tried this method:
I created a table tab_test which has these records:
TNAME                          TABTYPE    CLUSTERID
ABS_V4_P_ERROR_MESSAGES        TABLE
ABS_V4_P_ORG_PARAM             TABLE
ABS_V4_P_PARAMETER             TABLE
ABS_V4_P_SYS_PARAM             TABLE
ACCINTERFACE_PARAMETERS        TABLE
ACCOUNTS                       TABLE
ACCOUNT_EXTRACT_PERIODS        TABLE
ACCOUNT_EXTRACT_PERIODS#       TABLE
ACCOUNT_EXTRACT_PERIODS_1      TABLE                                                                                                                                                                                   Now I create this proc:
PROCEDURE FORMS_TEXT_DYN_SQL_TEST(p_where_cluase IN VARCHAR2, p_out_cursor OUT SYS_REFCURSOR) IS
v_stmt VARCHAR2(1000);
BEGIN
v_stmt := 'SELECT tname FROM tab_test WHERE tname LIKE ''%ABS_V4%'' AND tabtype = :x';
OPEN p_out_cursor FOR v_stmt using p_where_cluase;
END;I create this code block and run it:
declare
v_tname varchar2(200);
out_cursor sys_refcursor;
begin
forms_text_dyn_sql_test('TABLE', out_cursor );
LOOP
    fetch out_cursor INTO v_tname;
    exit when out_cursor%NOTFOUND;
    DBMS_OUTPUT.PUT_LINE(v_tname);
END LOOP;
end;
/I get correct output:
ABS_V4_P_ERROR_MESSAGES
ABS_V4_P_ORG_PARAM
ABS_V4_P_PARAMETER
ABS_V4_P_SYS_PARAMHowever, when I change the proc like this:
PROCEDURE FORMS_TEXT_DYN_SQL_TEST(p_where_cluase IN VARCHAR2, p_out_cursor OUT SYS_REFCURSOR) IS
v_stmt VARCHAR2(1000);
BEGIN
v_stmt := 'SELECT tname FROM tab_test WHERE tname LIKE ''%ABS_V4%'' AND :y';
OPEN p_out_cursor FOR v_stmt using p_where_cluase;
END;And run this code block:
declare
v_tname varchar2(200);
out_cursor sys_refcursor;
begin
forms_text_dyn_sql_test(' 1 = 1 ', out_cursor );
LOOP
    fetch out_cursor INTO v_tname;
    exit when out_cursor%NOTFOUND;
    DBMS_OUTPUT.PUT_LINE(v_tname);
END LOOP;
end;
/I get error:
[1]: (Error): ORA-00920: invalid relational operator ORA-06512: at "ABS.FORMS_TEXT_DYN_SQL_TEST", line 6 ORA-06512: at line 5Looks like you can only put column_name = :z, column_name = :y type values. You cannot it seems replace it with any WHERE CLAUSE????

"create columns with a where clause" in BMM

Hi All,
I have a question about OBI EE Meta data layer.
We have OBIEE sitting on a Data Warehouse with a star schema.
Our Fact table holds the following fields...
EmployeeID
CompanyID
Measure ID
Measure Name
MEASURE
This is what we have in the physical layer, BMM layer and the presentation layer but I want to be able to "create columns with a where clause"
So my fact table in the presentation layer would look like this
Amount Sold
Unit Cost
Company Amount
I thought I could just create a logical column with
Amount Sold = Measure
where Measure Name = "Amount Sold"
But I can't seem to find any to do that.
Any ideas?
Thanks

Or you simply map your measure in the Logical Table source as something like :
CASE WHEN MEASURE_NAME = 'Amount Sold' THEN MEASURE ELSE 0 END.
You need to decide if you want the where clause adding in the Logical table source (be careful you will need a seperate LTS for each measure, this will mean more than one SQL being fired when you have more than one measure in the report).
Or you go with this case method on each measure, be able to have all the measures mapped in 1 LTS, include all those in the report and they will be retrieved by one SQL block.

Using SAMPLE with a WHERE clause

(Oracle 9iR2)
I am using the SAMPLE keyword in a SELECT statement with a WHERE clause. Is the sample extracted before or after applying the WHERE clause?
For example, if I have the statement:
SELECT emp_id FROM employees SAMPLE(10)
WHERE salary > 50000;
Does the database first extract a 10% sample of all the rows and then apply the salary filter on the sample (meaning it's possible that no rows are returned even though there are rows with salary > 50000)? Or does it return a 10% sample of the rows that meet the filter condition (meaning rows will always be returned as long as there are rows with salary > 50000)?

The sample is extracted before the where clause:
SQL> drop table sample_test;
Table dropped.
SQL> create table sample_test (id number, salary number);
Table created.
SQL>
SQL> begin
2 for i in 1..99 loop
3 insert into sample_test values (i, 100);
4 end loop;
5
6 insert into sample_test values (100, 1000);
7 end;
8 /
PL/SQL procedure successfully completed.
SQL>
SQL> commit;
Commit complete.
SQL>
SQL> select * from sample_test sample(5)
2 where salary > 100;
no rows selected

Insert Row is visible in other VO with excluded where-clause

Hi,
i found something that makes me a little bit confusing about ADF BC...
TestCase:
Env:
JDev 10.1.2
HR Schema / Table "regions"
All Objects created ´by ADF BC Wizards
- ViewObject A and ViewObject B based on Entity "Regions"
- ViewObject A with WHERE-CLAUSE region_id=10 at DesignTime
- ViewObject B with NO WHERE-CLAUSE (select * from regions)
JUnitTest:
public void testInsertRow()
ViewObject viewA = fixture1.getApplicationModule().findViewObject("RegionsAView");
ViewObject viewB = fixture1.getApplicationModule().findViewObject("RegionsBView");
System.out.println("ViewA:"+viewA.getRowCount()); //-->0
assertTrue("Must 0 ! region_id = 10",viewA.getRowCount() == 0);
Row row = viewB.createRow();
row.setAttribute("RegionId",new Number(99));
row.setAttribute("RegionName","NRW");
viewB.insertRow(row);
System.out.println("ViewB:"+viewB.getRowCount()); // -> 5
System.out.println("ViewA:"+viewA.getRowCount()); // -> 1
//this assert failt !
assertTrue("Must 0 ! where clause is region_id = 10 and insert region_id was 99 ",viewA.getRowCount() == 0);
Is this a bug or feature ? Can anybody explain that behavior ? thx :)
uli

See this article on my blog for an explanation and a code sample for how to gain more fine-grained control over this View Link Consistency feature...
http://radio.weblogs.com/0118231/2004/02/27.html#a242

REPORT with dynamic WHERE CLAUSE (run RDF or REP) ?

Hi:
When running a REPORT (myreport.rep) with dynamic where clause using a lexical parameter, I got this error:
REP-1439: Cannot compile .REP or .PLX file as it does not have source
If i run the report specifiying RDF extension (myreport.rdf) the report run successfully! Is this normal ?
If I specify RDF extension will Report Server COMPILE the report everytime I execute it ?
When using dynamic WHERE CLAUSE I will have to run RDF files instead of REP ?
I'm running Reports 9i under Linux, with IDS under Windows.
Waiting Help
Joao Oliveira

It sounds like you are building the .rep files on one platform (windows) and running them on another (linux). The reason that the .rdf file continues to work is that Reports recompiles the PL/SQL within the report when you move from one platform to another or change schemas. .rep files can't be re-compiled in this way so you need to ensure they are compiled successfully when converting them.
You need to convert from .rdf to .rep on the platform that you are intending to run on. Try running rwconverter on the linux platform with "compile_all=yes" to produce the .rep file and running that .rep file.

Performance hit using "where" clause in the query

Hi All,
I am facing a huge performance hit in the java code when using "where" clause in queries. Following are the details:
1. SELECT * FROM Employee
2. SELECT * FROM Employee where employeeid in (26,200,330,571,618,945)
There is no difference in Query Execution Time for both queries.
Business Logic Time is huge in second case as compared to first one (ratio - 1:20).
Rows returned are more in first case as compared to second case.(ratio - 1:4)
Business Logic is same for both the cases where I iterate through the ResultSet, get the objects and set them in a data structure.
Does anybody know the reason of unexpected time difference for the business logic in the second case?

Since you're mentioning clustering your index, I'll assume you are using Oracle. Knowing what database you are using makes it a lot easier to suggest things.
Since you are using Oracle, you can get the database to tell you what execution plan it is using for each of the 2 SQL statements, and figure out why they have similar times (if they do).
First, you need to be able to run SQL*Plus; that comes as part of a standard database installation and as part of the Oracle client installation - getting it set up and running is outside the scope of this forum.
Second, you may need your DBA to enable autotracing, if it's not already:
http://asktom.oracle.com/~tkyte/article1/autotrace.html
http://www.samoratech.com/tips/swenableautotrace.htm
Once it's all set up, you can log in to your database using sql*plus, issue "SET AUTOTRACE ON", issue queries and get execution plan information back.
For example:
SQL> set autotrace on
SQL> select count(*) from it.ticket where ticket_number between 10 and 20;
COUNT(*)
        11
Execution Plan
Plan hash value: 2983758974
| Id | Operation         | Name       | Rows | Bytes | Cost (%CPU)| Time     |
|   0 | SELECT STATEMENT |            |     1 |     4 |     1   (0)| 00:00:01 |
|   1 | SORT AGGREGATE   |            |     1 |     4 |            |          |
|* 2 |   INDEX RANGE SCAN| TICKET_N10 |    12 |    48 |     1   (0)| 00:00:01 |
Predicate Information (identified by operation id):
   2 - access("TICKET_NUMBER">=10 AND "TICKET_NUMBER"<=20)
Statistics
          0 recursive calls
          0 db block gets
          1 consistent gets
          0 physical reads
          0 redo size
        515 bytes sent via SQL*Net to client
        469 bytes received via SQL*Net from client
          2 SQL*Net roundtrips to/from client
          0 sorts (memory)
          0 sorts (disk)
          1 rows processed
SQL> This tells me that this query used an INDEX RANGE SCAN on index TICKET_N1; the query can't do much better than that logically... In fact, the statistic "1 consistent gets" tells me that Oracle had to examine only one data block to get the answer, also can't do better than that. the statistic, "0 physical reads" tells me that the 1 data block used was already cached in Oracle's memory.
the above is from Oracle 10g; autotrace is available back to at least 8i, but they've been adding information to the output with each release.
If you have questions about sql_plus, check the forums at asktom.oracle.com or http://forums.oracle.com/forums/category.jspa?categoryID=18
since sql*plus is not a JDBC thing...
Oh, and sql*plus can also give you easier access to timing information, with "set timing on".

Query SQL datastore with XML where clause source

Hope I am in the right place. New to Bus Obj Data Services Designer....I have cerated xml schemas, added it to the page as an xml source in. Mapped a test xml file and all is well there. I have added a query that grabs the xml.
I need to then query the MS SQL datastore ans use the data form the xml query as the where clause. How is this done? Or do I put a query on the datastore for all the data in a table then do anotehr query filtering one with the other? seems like that would be rather heavy and low performance. The results will then be sent back out as xml (schema and test file already set up as an xml out)
Thanks!

Thanks for the tips.
I'm trying to implement this option, using your ViewDefHelper.
I´m running into a problem though. After I create my dynamic View Object using a ViewDef, I need to create some view links.
So I get the AttributeDefs of the columns (source, and destination) from the method findAttributeDef (which is the whole purpose of performance in my post). This method is returning the correct Attribute Def, but when I create the view Link with the method createViewLinkBetweenViewObjects(java.lang.String vlName,
java.lang.String accessorName,
ViewObject master,
AttributeDef[] srcAttrs,
ViewObject detail,
AttributeDef[] destAttrs,
java.lang.String assocClause)
My destination query is generating the where clause as:
null = ?
Any Ideas what I'm doing wrong ?
Thanks again.
John.

Dynamic itab with Dynamic Where clause

Hi, Dear All,
Can someone provide a code extract for Dynamic where clause, i had already done with dynamic itab for a given set of fields, and i need to add where clause dynamically for a given field for a given range of values.
select (i_fields) into table <dyn_table>
                  from (p_table)
                 where (v_where).
In the above except the where clause, everything is done. Please help me.
with best regards
Mahesh

Hi,
here is the code extract for your reference.Pl. correct me.
with regards
REPORT Z_DYN_ITAB                              .
TYPE-POOLS: SLIS.
DATA:NAME(100) TYPE C.
TYPES: BEGIN OF ITAB_TYPE,
        WORD(20),
      END   OF ITAB_TYPE.
DATA: ITAB TYPE STANDARD TABLE OF ITAB_TYPE WITH HEADER LINE.
DATA: vg_fields(255) TYPE c,
      i_fields LIKE TABLE OF vg_fields.
DATA: it_fcat TYPE slis_t_fieldcat_alv,
      is_fcat LIKE LINE OF it_fcat,
      ls_layout TYPE slis_layout_alv.
DATA: it_fieldcat TYPE lvc_t_fcat,
      is_fieldcat LIKE LINE OF it_fieldcat.
field-symbols: <dyn_table> type standard table,
               <dyn_wa>,
               <dyn_field>.
data: dy_table type ref to data.
data: dy_line type ref to data,
      xfc type lvc_s_fcat.
DATA: v_where TYPE string, " Variable for storing where clause.
      v_dynamic(18) TYPE c, "variable to store select option datatype
      o_field TYPE REF TO cx_root," object to catch exception
      text TYPE string. "string variable to store exception text.
CONSTANTS: c_var(15) TYPE c VALUE ' IN S_RANGE'.
selection-screen begin of block b1 with frame.
parameters: p_table(30) type c default 'T001',
            p_name(100) type c,
            p_field(10) TYPE c. " Parameter to capture field name.
SELECT-OPTIONS: s_range FOR v_dynamic. " Select-option for range.
selection-screen end of block b1.
start-of-selection.
NAME = p_name.
SPLIT NAME AT ',' INTO TABLE ITAB.
LOOP AT ITAB.
is_fcat-fieldname = itab-word.
is_fcat-tabname = p_table.
APPEND is_fcat to it_fcat.
ENDLOOP.
LOOP AT it_fcat INTO is_fcat.
is_fieldcat-fieldname = is_fcat-fieldname.
is_fieldcat-tabname = is_fcat-tabname.
APPEND is_fieldcat TO it_fieldcat.
CONCATENATE is_fieldcat-tabname is_fieldcat-fieldname INTO
        vg_fields SEPARATED BY '~'.
APPEND vg_fields TO i_fields.
ENDLOOP.
perform create_dynamic_itab.
perform get_data.
Create dynamic internal table and assign to FS
form create_dynamic_itab.
call method cl_alv_table_create=>create_dynamic_table
exporting
it_fieldcatalog = it_fieldcat
importing
ep_table = dy_table.
assign dy_table->* to <dyn_table>.
Create dynamic work area and assign to FS
create data dy_line like line of <dyn_table>.
assign dy_line->* to <dyn_wa>.
endform.
form get_data.
Select Data from table.
CONCATENATE p_field c_var INTO v_where.
TRY.
select (i_fields) into table <dyn_table>
                  from (p_table)
                 where (v_where).
if sy-dbcnt = 0.
write : text-t02.
endif.
*Write out data from table.
Loop at <dyn_table> into <dyn_wa>.
do.
assign component sy-index of structure <dyn_wa> to <dyn_field>.
if sy-subrc <> 0.
exit.
endif.
if sy-index = 1.
write:/ <dyn_field>.
else.
write: <dyn_field>.
endif.
enddo.
endloop.
Exception Catching.
CATCH cx_root INTO o_field.
text = o_field->get_text( ).
Calling Function to give information message regarding Exception
CALL FUNCTION 'POPUP_TO_INFORM'
EXPORTING
titel = text-t03
txt1 = text
txt2 = text-t04.
TXT3 = ' '
TXT4 = ' '
LEAVE TO LIST-PROCESSING.
ENDTRY.
endform.

Outer join with a WHERE clause

hi, this is driving me round the bend. I thought i was ok with sql joins and the like, but im really struggling on this one.
I have table A (for argument sake tblRegion) and table B (tblBranch). if i want to return everything i would do something like this:-
select * from tblRegion join tblBranch on tblRegion.id=tblBranch.reg_id.
If i wanted to return all regions even if they do have a branch associated i would do something like this:-
select * from tblRegion left outer join tblBranch on tblRegion.id=tblBranch.reg_id.
My problem is, I want to return all Regions and use a WHERE clause to narrow the Branch results.
I've tried something like this but it doesnt work
select * from tblRegion left outer join tblBranch on tblRegion.id=tblBranch.reg_id WHERE branch.name like 'LONDON%'
Is there anyway i can return all branches where name is like LONDON, and STILL return all Region if there is no data for that Region?
Do i need to use a nested select statement maybe?
Many thanks in anticipation somebody will save me from my current madness. David

Check this link. It explains the most common problem/coding error while using outer join.
http://www.orafaq.com/node/855
Effectively in your case you need the following condition in your where clause
where nvl(branch.name,'LONDON') like 'LONDON%'
Hope that helps.
Regards
Raj

Loop with dynamic where clause

hi,
In a program we need to use a loop into an internal table with a where condition is decided on run time.
"Select" from database table is using dynamic where clause and it is working fine.
For the loop part does any one have any idea on this? if we can have dynamic where clause in a loop into internal tab?
Thanks

Hi,
You can't use 'loop where' dynamically .
Have you already tried to use Ranges and CHECK statement?
Like:
First mount range options;
Loop at <table>.
Check using the range.
endloop.
Maybe it's works for you !!
Marcelo Ramos

Binding parameters to dynamic VO with PL/SQL API call with no where clause

Hi all,
I am required to change exisiting queries to queries with "bind" parameters. Some of our VOs are dynamic and with PL/SQL api calls like below:
String stmtStr = "selectXXX_API_UTILS.chk_urg_since_last_prg("
+ projectId + "," + PPCId + ",'XYZ_TASKS'," + taskId+ ")"
+" from dual ";
ViewObject tmpVO = transaction.createViewObjectFromQueryStmt(stmtStr);
In this regard, I am unsure how to bind the PPCId and taskId parameters to the VO. setWhereClauseParams() would not work here as there is now where clause.
Thanks in advance,
Srini

In case of preparedStatement, we mention bind parameters to be passed using "?". Then we pass paramters sequencially.
But in the your case following is enough:
int projectId = 100 ; (hardcoding values for example)
int taskId = 500 ;
String stmtStr = "selectXXX_API_UTILS.chk_urg_since_last_prg("
+ projectId + "," + PPCId + ",'XYZ_TASKS'," + taskId+ ")"
+" from dual ";
-Anand

Finding undisplayable characters in a field with a WHERE clause

Oracle_Rdb7 for_OpenVMS Release 7.0.1. I have a field in my table that has been loaded with data that contains "return/line feed" characters (I'm not sure of the ASCII hex representation) which can't be displayed and show up as "." on my screen. How can I code a WHERE clause that will find these characters, e.g., SELECT * FROM TABLE1 WHERE NAME CONTAINING <here is where I want to put the ASCII hex value for a "return/line feed" character>? Also, how could I get the actual ASCII hex code for these characters to display, e.g., "31" instead of "1"?

a 'return' is CHR(10) So you can:
select * from my_table where text_fld like '%'||CHR(10)||'%'

Most efficient way to get row count with a where clause

Have not found a definitive answer on how to do this. Is there a better way to get a row count from a large table that needs to satisfy a where clause like so:
SELECT COUNT(*) FROM BigTable WHERE TypeName = 'ABC'
I have seen several posts suggesting something like
SELECT COUNT(*) FROM BigTable(NOLOCK);
and
selectOBJECT_NAME(object_id),row_count from sys.dm_db_partition_stats
whereOBJECT_NAME(object_id)='BigTable'but I need the row count that satisfies my where clause (i.e.: WHERE TypeName = 'ABC')

It needs index to improve the performance.
- create index on typename column
- create a indexed view to do the count in advance
-partition on type name (although it's unlikely) then get count from the system tables...
all those 3 solutions are about indexing.
Regards
John Huang, MVP-SQL, MCM-SQL, http://www.sqlnotes.info

Can you help me with the WHERE clause? Any issue with this code block?

Hi,
I am looking for blogs or any document on how to implement a Lookup during data loads in BW.
The problem is that all those that I find on this site are mostly about u201CLookup in XIu201D
I have read bits and hints on different postings on this site and this is the best I could do for the scenario shown below:
I need to add the Chars: Char1, Char2, Char3 to CubeX
so that I can display the data for these 3 Chars in a report.
Char1 and Char2 are filled in ODS12 and Char3 is filled in Cube3.
Can you help me write a routine so that while loading data to CubeX, it will read Char1, Char2, Char3 from ODS12 and Cube3; and make them available in CubeX?
This is my attempt so far: continue:
u201C----
select /bic/Char1 from /bic/ODS12
select /bic/Char2 from /bic/ODS12
select /bic/Char3 from /bic/Cube3
into result
WHERE u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026u2026
and objvers = 'A' .
end select.
u201C----
1. What should be my WHERE clause?
2. In the select statement, by studying samples on this where should there be an u201CMu201D in front of the ODS? e.g. select /bic/Char1 from /bic/MODS12
If so, what is the significance and the other options?
3. Should this code be in the Start routing of CubeX? Or any other possible location? Also, in the start routine does it matter where within the start routine this code needs to be placed?
4. Can you please fix this code and add any lines I may be missing?
Is what I am looking to do refer to as u201Cdoing a lookupu201D?
Thanks

ok

Performance with unspecific where clause

Similar Messages

Maybe you are looking for