Oracle database supports fetch reblocking with the HS_RPC_FETCH_REBLOCKING parameter.

When the value of this parameter is set to ON (the default), the array size for SELECT statements is determined by the HS_RPC_FETCH_SIZE value. The HS_RPC_FETCH_SIZE parameter defines the number of bytes sent with each buffer from the gateway to the Oracle database. The buffer may contain one or more qualified rows from the DRDA server. This feature can provide significant performance enhancements, depending on your application design, installation type, and workload.

The array size between the client and the Oracle database is determined by the Oracle application. Refer to Oracle Database Gateway Installation and Configuration Guide for IBM AIX on POWER Systems (64-Bit), Linux x86-64, Oracle Solaris on SPARC (64-Bit), Oracle Solaris on x86-64 (64-Bit) and HP-UX Itanium or Oracle Database Gateway Installation and Configuration Guide for Microsoft Windows , depending on your platform, for more information.

The following are special considerations for using the procedural feature with the gateway:

• PL/SQL records cannot be passed as parameters when invoking a DB2 stored procedure.

• The gateway supports the GENERAL and DB2SQL linkage conventions of DB2 stored procedures. Both linkage conventions require that the parameters that are passed to and from the DB2 stored procedure cannot be null.

The Oracle Database Gateway for DRDA provides support for stored procedures that return result sets. By default, all stored procedures and functions do not return a result set to the user. To enable result sets, set the HS_FDS_RESULTSET_SUPPORT parameter in the initialization parameter file.

When accessing stored procedures with result sets through the Oracle Database Gateway for DRDA, you will be in the sequential mode of Heterogeneous Services. The gateway returns the following information to Heterogeneous Services during procedure description:

• All the input arguments of the remote stored procedure

• None of the output arguments

• One out argument of type ref cursor (corresponding to the first result set returned by the stored procedure)

Client programs have to use the virtual package function DBMS_HS_RESULT_SET.GET_NEXT_RESULT_SET to get the ref cursor for subsequent result sets. The last result set returned is the out argument from the procedure.

The limitations of accessing result sets are as follows:

• Result sets returned by a remote stored procedure have to be retrieved in the order in which they were placed on the wire.

• On execution of a stored procedure, all result sets returned by a previously executed stored procedure will be closed, regardless of whether the data has been completely retrieved or not.

In the following example, the UDB stored procedure is executed to fetch the contents of the EMP and DEPT tables from UDB:

CREATE PROCEDURE REFCURPROC (IN STRIN VARCHAR(255), OUT STROUT VARCHAR(255) ) 
  RESULT SETS 3  LANGUAGE SQL 
BEGIN
  DECLARE TEMP CHAR (20); 
 DECLARE C1 CURSOR WITH RETURN TO CALLER FOR 
      SELECT * FROM TKHOEMP; 
  DECLARE C2 CURSOR WITH RETURN TO CALLER FOR 
     SELECT * FROM TKHODEPT; 
 OPEN C1; 
 OPEN C2; 
 SET STROUT = STRIN; 
 END

OCIEnv *ENVH;
OCISvcCtx *SVCH;
OCIStmt *STMH;
OCIError *ERRH;
OCIBind *BNDH[3];
OraText arg1[20];
OraText arg2[255];
OCIResult *rset;
OCIStmt *rstmt;
ub2 rcode[3];
ub2 rlens[3];
sb2 inds[3];
OraText *stmt = (OraText *) "begin refcurproc@UDB(:1,:2,:3); end;";
OraText *n_rs_stm = (OraText *)
  "begin :ret := DBMS_HS_RESULT_SET.GET_NEXT_RESULT_SET@UDB; end;";
 
/* Prepare procedure call statement */
 
/* Handle Initialization code skipped */
OCIStmtPrepare(STMH, ERRH, stmt, strlen(stmt), OCI_NTV_SYNTAX, OCI_DEFAULT);
 
/* Bind procedure arguments */
inds[0] = 0;
strcpy((char *) arg1, "Hello World");
rlens[0] = strlen(arg1);
OCIBindByPos(STMH, &BNDH[0], ERRH, 1, (dvoid *) arg1, 20, SQLT_CHR,
             (dvoid *) &(inds[0]), &(rlens[0]), &(rcode[0]), 0, (ub4 *) 0, 
             OCI_DEFAULT);
 
inds[1] = -1;
OCIBindByPos(STMH, &BNDH[1], ERRH, 1, (dvoid *) arg2, 20, SQLT_CHR,
             (dvoid *) &(inds[1]), &(rlens[1]), &(rcode[1]), 0, (ub4 *) 0, 
             OCI_DEFAULT);
 
inds[2] = 0;
rlens[2] = 0;
OCIDescriptorAlloc(ENVH, (dvoid **) &rset, OCI_DTYPE_RSET, 0, (dvoid **) 0);
OCIBindByPos(STMH, &BNDH[2], ERRH, 2, (dvoid *) rset, 0, SQLT_RSET,
             (dvoid *) &(inds[2]), &(rlens[2]), &(rcode[2]),
             0, (ub4 *) 0, OCI_DEFAULT);
 
/* Execute procedure */
OCIStmtExecute(SVCH, STMH, ERRH, 1, 0, (CONST OCISnapshot *) 0,
               (OCISnapshot *) 0, OCI_DEFAULT);
 
/* Convert result set to statement handle */
OCIResultSetToStmt(rset, ERRH);
rstmt = (OCIStmt *) rset;
 
/* After this the user can fetch from rstmt */
/* Issue get_next_result_set call to get handle to next_result set */
/* Prepare Get next result set procedure call */
 
OCIStmtPrepare(STMH, ERRH, n_rs_stm, strlen(n_rs_stm), OCI_NTV_SYNTAX,
               OCI_DEFAULT);
 
/* Bind return value */
OCIBindByPos(STMH, &BNDH[1], ERRH, 1, (dvoid *) rset, 0, SQLT_RSET,
             (dvoid *) &(inds[1]), &(rlens[1]), &(rcode[1]),
             0, (ub4 *) 0, OCI_DEFAULT);
 
/* Execute statement to get next result set*/
OCIStmtExecute(SVCH, STMH, ERRH, 1, 0, (CONST OCISnapshot *) 0,
               (OCISnapshot *) 0, OCI_DEFAULT);
 
/* Convert next result set to statement handle */
OCIResultSetToStmt(rset, ERRH);
rstmt = (OCIStmt *) rset;
 
/* Now rstmt will point to the second result set returned by the
remote stored procedure */
 
/* Repeat execution of get_next_result_set to get the output arguments */

create or replace package rcpackage is  type RCTYPE is ref cursor;end rcpackage;/
declare
  rc1 rcpackage.rctype;
  rec1 loc_emp%rowtype;
  rc2 rcpackage.rctype;
  rec2 loc_dept%rowtype;
  rc3 rcpackage.rctype;
  rec3 outargs%rowtype;
  out_arg varchar2(255);
 
begin
 
  -- Execute procedure
  out_arg := null;
  refcurproc@UDB('Hello World', out_arg, rc1);
 
  -- Fetch 20 rows from the remote emp table and insert them into loc_emp
  for i in 1 .. 20 loop
    fetch rc1 into rec1;
    insert into loc_emp (rec1.empno, rec1.ename, rec1.job,
    rec1.mgr, rec1.hiredate, rec1.sal, rec1.comm, rec1.deptno);
  end loop;
 
  -- Close ref cursor
  close rc1;
 
  -- Get the next result set returned by the stored procedure
  rc2 := dbms_hs_result_set.get_next_result_set@UDB;
 
  -- Fetch 5 rows from the remote dept table and insert them into loc_dept
  for i in 1 .. 5 loop
    fetch rc2 into rec2;
    insert into loc_dept values (rec2.deptno, rec2.dname, rec2.loc);
  end loop;
 
  --Close ref cursor
  close rc2;
 
  -- Get the output arguments from the remote stored procedure
  -- Since we are in sequential mode, they will be returned in the
  -- form of a result set
  rc3 := dbms_hs_result_set.get_next_result_set@UDB;
 
  -- Fetch them and insert them into the outarguments table
  fetch rc3 into rec3;
  insert into outargs (rec3.outarg, rec3.retval);
 
  -- Close ref cursor
  close rc3;
 
end;
/

Oracle database automatically forwards compatible SQL functions to the DRDA database, where SQL constructs with the same syntax and meaning are on both Oracle database and the DRDA database. These SQL constructs are forwarded unmodified. All of the compatible functions are column functions. Functions that are not compatible are either translated to an equivalent DRDA SQL function or are compensated (post-processed) by Oracle database after the data is returned from the DRDA database.

Translated functions have the same meaning but different names between the Oracle database and the DRDA database. But all applications must use the Oracle function name. These SQL constructs that are supported with different syntax (different function names) by the DRDA database, are automatically translated by the Oracle database and then forwarded to the DRDA database. Oracle database changes the function name before sending it to the DRDA database, in a manner that is transparent to your application.

Some advanced SQL constructs that are supported by Oracle database may not be supported in the same manner, by the DRDA database. Compensated functions are those SQL functions that are either not recognized by the DRDA server or are recognized by the DRDA server but the semantics of the function are interpreted differently when comparing the DRDA server with the Oracle database. If a SELECT statement containing one of these functions is passed from the Oracle database to the gateway, then the gateway removes the function before passing the SQL statement to the DRDA server. The gateway passes the selected DRDA database rows to Oracle database. Oracle database applies the function.

SELECT COS(X_COOR) FROM TABLE_X;

SELECT X_COOR FROM TABLE_X;

Some SQL functions that are normally compensated may also be overridden, through the Native Semantics facility. If a SQL function has been enabled for Native Semantics, then the function may be passed on to the DRDA database for processing, instead of being compensated. The SQL function is then processed natively in the DRDA database. Refer to "Native Semantics" for more information.

Table 4-1 describes how Oracle database and the gateway handle SQL functions for a DB2 UDB for z/OS.

ABS	ACOS
ASCII	ASIN
ATAN	ATAN2
BITAND	CAST
CEIL	CHR
CONVERT	COS
COSH	COUNTCOL
DECODE	DUMP
EXP	FLOOR
GREATEST	HEXOTRAW
INITCAP	INSTR
INSTRB	LEAST
LENGTH	LENGTHB
LN	LOG
LOWER	LPAD
LTRIM	MOD
NLS_INITCAP	NLS_UPPER
NLS_LOWER	NLSSORT
NVL2	POWER
RAWTOHEX	REPLACE
REVERSE	ROUND
RPAD	RTRIM
SIGN	SIN
SINH	SQRT
STDDEV	SUBSTR
SUBSTRB	TAN
TANH	TO_NUMBER
TRANSLATE	TRIM
TRUNC	UPPER
VARIANCE	VSIZE

The following SQL functions are enabled (ON) by default:

• GROUPBY

• HAVING

• ORDERBY

• WHERE

ORDERBY controls sort order, which may differ at various sort locations. For example, with ORDERBY ON, a DB2 sort would be based on Extended Binary Coded Decimal Interchange Code (EBCDIC) sorting order, whereas with ORDERBY OFF, an Oracle database sort would be based on ASCII sorting order.

The other three functions, GROUPBY, HAVING, and WHERE, can take additional processing time. If you need to minimize the use of expensive resources, then you should choose the settings of these functions so that the processing is performed with cheaper resource. The above listed functions can also be disabled.

The WHERE and HAVING clauses are compatible for all versions of the DRDA server. This means that these clauses are passed unchanged to the DRDA server for processing. Whether clauses GROUP BY and ORDER BY are passed to the DRDA server, or compensated by Oracle database, is determined by the Native Semantics Parameters (see the previous section).

The set operators UNION and UNION ALL are compatible for all versions of the DRDA server, meaning that they are passed unchanged to the DRDA server for processing. The set operators INTERSECT and MINUS are compensated on all versions of the DRDA server except DB2/UDB. For DB2/UDB, INTERSECT is compatible and MINUS is translated to EXCEPT.

The gateway performs all character string comparisons, concatenations, and sorts using the data type of the referenced columns, and determines the validity of character string values passed by applications using the gateway. The gateway automatically converts character strings from one data type to another and converts between character strings and dates when needed.

Frequently, DRDA databases are designed to hold non-character binary data in character columns. Applications executed on DRDA systems can generally store and retrieve data as though it contained character data. However, when an application accessing this data runs in an environment that uses a different character set, inaccurate data may be returned.

With the gateway running on the host, character data retrieved from a DB2 UDB for iSeries or DB2 UDB for z/OS host is translated from EBCDIC to ASCII. When character data is sent to DB2 UDB for iSeries or DB2 UDB for z/OS from the host, ASCII data is translated to EBCDIC. When the characters are binary data in a character column, this translation causes the application to receive incorrect information or errors. To resolve these errors, character columns on DB2 UDB for iSeries or DB2 UDB for z/OS that hold non-character data must be created with the FOR BIT DATA option. In the application, the character columns holding non-character data should be processed using the Oracle data types RAW and LONG RAW. The DESCRIBE information for a character column defined with FOR BIT DATA on the host always indicates RAW or LONG RAW.

The gateway binds character string data values from host variables as fixed-length character strings. The bind length is the length of the character string data value. The gateway performs this conversion on every bind.

The DRDA VARCHAR data type can be between 1 and 32767 characters in length if the Oracle database is configured to use maximum VARCHAR2 size of 32767. Otherwise, the limit is 4000. If the DRDA VARCHAR data type is greater than the Oracle configured VARCHAR2 limit size, then it is converted to an Oracle LONG data type.

The DB2 VARCHAR data type can be no longer than 32767 bytes, which is much shorter than the maximum size for the Oracle LONG data type. If you define an Oracle LONG data type larger than 32767 bytes in length, then you receive an error message when it is mapped to the DB2 VARCHAR data type.

DB2 GRAPHIC data types store only double-byte string data. Sizes for DB2 GRAPHIC data types typically have maximum sizes that are half that of their character counterparts. For example, the maximum size of a CHAR may be 255 characters, whereas the maximum size of a GRAPHIC may be 127 characters.

Oracle database does not have a direct matching data type, and the gateway therefore converts between Oracle character data types to DB2 Graphic data types. Oracle database character data types may contain single, mixed, or double-byte character data. The gateway converts the string data into appropriate double-byte-only format depending upon whether the target DB2 column is a Graphic type and whether Gateway Initialization parameters are set to perform this conversion. For more configuration information, refer to Initialization Parameters.

The implementation of date and time data differs significantly in IBM DRDA databases and Oracle database. Oracle database has a single date data type, DATE, which can contain both calendar date and time of day information.

IBM DRDA databases support the following three distinct date and time data types:

DATE is the calendar date only.

TIME is the time of day only.

TIMESTAMP is a numerical value combining calendar date and time of day with microsecond resolution in the internal format of the IBM DRDA database.

SELECT * FROM EMP WHERE HIREDATE = '03-MAR-81'

SELECT * FROM EMP WHERE HIREDATE = TO_DATE('03-MAR-81')

SELECT * FROM EMP WHERE HIREDATE = TO_DATE(:1)

SELECT * FROM EMP WHERE HIREDATE = '1981-03-03'

date + number 
number + date 
date - number 
date1 - date2 

Date handling has two categories: 

• Two-digit year dates, which are treated as occurring 50 years before or 50 years after the year 2000.

• Four-digit year dates, which are not ambiguous with regard to the year 2000.

Use one of the following methods to enter twenty-first century dates:

• The TO_DATE function

  Use any date format including a four-character year field. Refer to the Oracle Database SQL Language Reference for the available date format string options.

  For example, TO_DATE('2008-07-23', 'YYYY-MM-DD') can be used in any SELECT, INSERT, UPDATE, or DELETE statement.

• The NLS_DATE_FORMAT parameter

  ALTER SESSION SET NLS_DATE_FORMAT should be used to set the date format used in SQL.

The following table lists the four patterns that can be used for the NLS_DATE_FORMAT in ALTER SESSION SET NLS_DATE_FORMAT:

The Oracle database default format of 'DD-MON-YY' is not permitted with DB2.

The following example demonstrates how to enter and select date values in the twenty-first century:

ALTER SESSION SET NLS_DATE_FORMAT = 'YYYY-MM-DD';
INSERT INTO EMP (HIREDATE) VALUES ('2008-07-23');
SELECT * FROM EMP WHERE HIREDATE = '2008-07-23';
UPDATE EMP SET HIREDATE = '2008-07-24'
   WHERE HIREDATE = '2008-07-23';
DELETE FROM EMP WHERE HIREDATE = '2008-07-24';

The Oracle TO_DATE function is preprocessed in SQL INSERT, UPDATE, DELETE, and SELECT WHERE clauses. TO_DATE functions in SELECT result lists are not preprocessed.

The TO_DATE function is often needed to provide values to update or compare with date columns. Therefore, the gateway replaces the information included in the TO_DATE clause with an acceptable value before the SQL statement is sent to DB2.

Except for the SELECT result list, all TO_DATE functions are preprocessed and turned into values that are the result of the TO_DATE function. Only TO_DATE(literal) or TO_DATE(:bind_variable) is permitted. Except in SELECT result lists, the TO_DATE(column_name) function format is not supported.

The preprocessing of the Oracle TO_DATE functions into simple values is useful in an INSERT VALUES clause because DB2 does not allow functions in the VALUES clause. In this case, DB2 receives a simple value in the VALUES list. All forms of the TO_DATE function (with one, two, or three operands) are supported.

IBM versions of the DRDA server perform automatic conversions to the numeric data type of the destination column (such as integer, double-precision floating point, or decimal). The user has no control over the data type conversion, and this conversion can be independent of the data type of the destination column in the database.

For example, if PRICE is an integer column of the PRODUCT table in an IBM DRDA database, then the update shown in the following example inaccurately sets the price of an ice cream cone to $1.00 because the IBM DRDA server automatically converts a floating point to an integer:

UPDATE PRODUCT 
SET PRICE = 1.50 
WHERE PRODUCT_NAME = 'ICE CREAM CONE    '; 

Because PRICE is an integer, the IBM DRDA server automatically converts the decimal data value of 1.50 to 1.

Oracle database supports the following four operands for the COUNT function:

• COUNT(*)

• COUNT(DISTINCT colname)

• COUNT(ALL colname)

• COUNT(colname)

Some DRDA servers do not support all forms of COUNT, specifically COUNT(colname) and COUNT(ALL colname). In those cases the COUNT function and its arguments are translated into COUNT(*). This may not yield the desired results, especially if the column being counted contains NULLs.

For those DRDA servers that do not support the above forms, it may be possible to achieve equivalent functionality by adding a WHERE clause. For example,

SELECT COUNT(colname) FROM table@dblink WHERE colname IS NOT NULL 

or

SELECT COUNT(ALL colname) FROM table@dblink WHERE colname IS NOT NULL

You can also use native semantics to indicate support for all COUNT functions with the following parameter in your gateway initialization file:

HS_FDS_CAPABILITY=(COUNTCOL=YES)

Refer to SQL Limitations for known DRDA servers that do not support all forms of COUNT.

A zoned decimal field is described as packed decimal on Oracle database. However, an Oracle application such as a Pro*C program can insert into a zoned decimal column using any supported Oracle numeric data type. The gateway converts this number into the most suitable data type. Data can be fetched from a DRDA database into any Oracle data type, provided that it does not result in a loss of information.

As noted above, SQL statements that are processed through the DBMS_HS_PASSTHROUGH.EXECUTE_IMMEDIATE function are not interpreted by the Oracle database. As a result, the Oracle database will not know if such statements are making any modifications to the DRDA server. This means that unless you keep the cached information of the Oracle database up to date after changes to the DRDA server, the database may continue to rely upon inaccurate or outdated information in subsequent queries within the same session.

An example of this occurs when you alter the structure of a table by either adding or removing a column. When an application references a table through the gateway (for example, when you perform a query on it), the Oracle database caches the table definition. Now, suppose that within the same session, the application subsequently alters the table's form, by using DBMS_HS_PASSTHROUGH.EXECUTE_IMMEDIATE to add a column. Then, the next reference to the table by the application will return the old column definitions of the table and will ignore the table's new column. This is because the Oracle database did not process the statement and, so, has no knowledge of the alteration. Because the database does not know of the alteration, it has no reason to requery the table form, and, so, it will use the already-cached form to handle any new queries.

In order for the Oracle database to acquire the new form of the table, the existing session with the gateway must be closed and a new session must be opened. This can be accomplished in either of two ways:

• By ending the application session with the Oracle database and starting a new session after modifications have been made to the DRDA server; or

• By running the ALTER SESSION CLOSE DATABASE LINK command after making any modifications to the DRDA server.

Either of the above actions will void the cached table definitions and will force the Oracle database to acquire new definitions on the next reference.

DECLARE
  CRS binary_integer;
  RET binary_integer;
  VAL VARCHAR2(10)
BEGIN
  CRS:=DBMS_HS_PASSTHROUGH.OPEN_CURSOR@gtwlink;
  DBMS_HS_PASSTHROUGH.PARSE@gtwlink(CRS,'SELECT NAME FROM PT_TABLE');
BEGIN
  RET:=0;
  WHILE (TRUE)
  LOOP
    RET:=DBMS_HS_PASSTHROUGH.FETCH_ROW@gtwlink(CRS,FALSE);
    DBMS_HS_PASSTHROUGH.GET_VALUE@gtwlink(CRS,1,VAL);
    INSERT INTO PT_TABLE_LOCAL VALUES(VAL);
  END LOOP;
  EXCEPTION
    WHEN NO_DATA_FOUND THEN
      BEGIN
        DBMS_OUTPUT.PUT_LINE('END OF FETCH');
        DBMS_HS_PASSTHROUGH.CLOSE_CURSOR@gtwlink(CRS);
      END;
    END;  
END;
/ 

The gateway data dictionary views provide users with an Oracle-like interface to the contents and use of the DRDA database. Oracle products require some of these views. The gateway supports the DB2 UDB for z/OS, DB2 UDB for iSeries, and DB2/UDB catalog views.

You can query the gateway data dictionary views to see the objects in the DRDA database and to determine the authorized users of the DRDA database. The Oracle Database Gateway for DRDA supports many Oracle catalog views. Refer to Oracle DB2 Data Dictionary Views for descriptions of Oracle DB2 catalog views. These views are completely compatible with the gateway.

Each DRDA database has its own catalog tables and views, which you might find useful. Refer to the appropriate IBM documentation for descriptions of these catalogs.