Support MC Press - Visit Our Sponsors

Forums Sponsor

Popular Forums

POPULAR FORUMS Replacing PLISTS in FREE code (764 views) DBinfo - AS/400 Free Database Editor (716 views) Errors in joblog when running "Hello Excel" sample code (538 views) Considering the World Beyond WebSphere (502 views) ATtn of Joe Pluta! EGL for IBM i. Where to find the SAVF with IRUI Library for i syst (451 views) Tribute to Craig Johnson (306 views) Firewall issue with FTPS from iSeries to WINDOWS (272 views) Announcing VRTSAV, a new open-source utility from the Easy400.net site (127 views) Downloads (102 views)

Forums

 

 

   Search Sponsor

POPULAR SEARCHES

Search

New capabilities allow designers and programmers to continue to produce new and exciting applications.

Because DB2 for i5/OS is embedded within i5/OS, application designers and programmers likely expected that with the V6R1 release of i5/OS there would be plenty of database enhancements included. They weren't disappointed, because the developers of DB2 for i5/OS have been busy developing many new functions and enhancements that should be of interest to anyone that works with the database. (Isn't that everyone?)

I certainly will not be able to cover all the new functions and enhancements. Nor will I be able to describe any in full detail. As a way to discuss this lineup of new functions and enhancements, they will be broken up into several categories. These will include SQL syntax, new catalog statistic views, application development, SQL Query Engine (SQE) and performance, increased limits, and the System i Navigator database tools.

Starting Off with the SQL Syntax

Query users will find the VALUES clause helpful when they want to add literals or variable values to their result set without creating another table first.

SELECT      VAL1, VAL2,

            WEEK(SALES_DATE) AS WEEK,

            DAYOFWEEK(SALES_DATE) AS DAY_WEEK,

            MONTH(SALES_DATE) AS MONTH,

            REGION,

            SALES AS UNITS_SOLD

FROM PRODAPPL.SALES, 

            (VALUES('GROUP 1','GROUP 2') ) AS X (VAL1,VAL2)

Also, this release brings true full outer join support, which is useful for joining tables where some rows from the joined tables may not have any rows.

The ORDER BY and the FETCH FIRST n ROWS ONLY clauses will now be allowed on any subselect. For example, if an application needs to union tables together but only wants the top 10 sales from different regions tables, this support makes this possible, as shown below:

(SELECT * FROM PRODAPPL.SALES_EAST

      ORDER BY SALES DESC FETCH FIRST 10 ROWS ONLY)

UNION ALL

(SELECT * FROM PRODAPPL.SALES_CENT

      ORDER BY SALES DESC FETCH FIRST 10 ROWS ONLY)

UNION ALL

(SELECT * FROM PRODAPPL.SALES_WEST

      ORDER BY SALES DESC FETCH FIRST 10 ROWS ONLY)

Another nice query enhancement is the INSERT in the FROM clause. This will help with cases where the value is generated by DB2 for i5/OS. It allows the programmer to access the result rows of the INSERT statement as part of the SELECT statement.

SELECT *

FROM FINAL TABLE   ( INSERT INTO PRODAPPL.SALES_COPY ( SALES_DATE,

                                    SALES_PERSON, REGION, SALES )

                             SELECT * FROM PRODAPPL.SALES )

ORDER BY INPUT SEQUENCE;

In addition, grouping sets and super groups were introduced, which allows the programmer to do multiple groupings and include the groups in the final result set. In prior releases, there was no way to do this except to run multiple queries.

SELECT WEEK(SALES_DATE) AS WEEK,

                  DAYOFWEEK(SALES_DATE) AS DAY_OF_WEEK,

                  SALES_PERSON,

                  SUM(SALES) AS NUMBER_OF_UNITS_SOLD

     FROM PRODAPPL.SALES

     WHERE WEEK(SALES_DATE) = 10

     GROUP BY GROUPING SETS( (WEEK(SALES_DATE), SALES_PERSON),

                  (DAYOFWEEK(SALES_DATE), SALES_PERSON) )

     ORDER BY WEEK, DAY_OF_WEEK, SALES_PERSON

Another clause that has been added is SKIP LOCKED DATA, which allows the SQL program to continue retrieving rows from the target table when doing a query, but skip over the rows that are locked. Previously, the query would have timed out or waited for the lock to be released This is especially useful for queuing applications where rows can be processed by multiple applications in any order. 

New extended indicator variable values provide a way to specify which columns in an UPDATE statement should actually be updated. Unchanged values can be tagged with a new indicator value to remove them from the update. Previously, the programmer would have had to code this with multiple update statements, and the statements would have been individually optimized.

In addition, there are several new scalar functions in V6R1:

Also, a new data type named DECFLOAT has been added. A form of floating point, it keeps precision accurately like a decimal number. It is an IEEE 754R number with a decimal point. Special values even allow the program to store infinity! This data type is fully supported in SQL statements. Note, however, that only the ILE/C compiler supports DECFLOAT in this release.

Also, new data type syntax has been added to support the standard way to define UTF-16 columns.

Here's an example:

CREATE TABLE myUTF16.table (

           C1 NCHAR(10),

           C2 NVARCHAR(20),

           C3 NCLOB(1M) );

Another nice addition is the ability to create a column that has the IMPLICITLY HIDDEN attribute. This allows programmers to define and use the column in their applications, but it effectively "hides" the column from query applications if users or programs do not explicitly specify the column.

There is also now a way to define a timestamp column that will be automatically updated whenever a row changes by using the FOR EACH ROW ON UPDATE AS ROW CHANGE TIMESTAMP clause. This clause could be used in conjunction with the IMPLICITLY HIDDEN clause to have a hidden column that is updated with the latest change timestamp for each row.

CREATE TABLE PRODAPPL.TABLE1 (

      C1 INT,

      C2 VARCHAR (20),

      C3 TIMESTAMP IMPLICITLY HIDDEN;

SELECT * FROM PRODAPPL.TABLE1;

The result of this query will display only columns C1 and C2. C3 will be hidden from the query user's result set. But if the query is changed to include C1, C2, and C3, then it will return all columns.

SELECT C1, C2, C3 FROM PRODAPPL.TABLE1;

A significant enhancement is the ability to use expressions when creating an index. The CREATE INDEX statement now allows the use of expression notation to define the key columns. For example, SALARY + BONUS could be created as a key column in an index, which is commonly called a derived key. This will help give the optimizer a more useful key to consider when an application's query is looking for these columns combined. This support is similar to a derived key logical file. Note that an RPG program can use an index with selection using native I/O, but the program cannot use it in an SQL statement.

CREATE INDEX PRODAPPL.TOTAL_SALARY ON PRODAPPL.EMPLOYEE (SALARY + BONUS ASC)

A new capability that will be a real time-saver is the ability to use the ALTER statement to alter a function definition. Previously, the function would have to have been dropped and re-created with the new definition, and in the process any authorizations that were defined on the function were lost.

Sometimes, 50 characters were not enough text to describe constraint relationships for a programmer, so the COMMENT ON statement has been extended to constraints. The LABEL ON statement has also been extended to more SQL objects.

For programmers interested in having only one set of SQL syntax for their SQL statements while at the same time supporting multiple database targets or programmers wanting to port their SQL statement from another database to DB2 for i5/OS, many DDL statements and clauses that are unnecessary on DB2 for i5/OS will now be ignored.

Also, trigger support has been enhanced to allow the use of an INSTEAD OF trigger to be run now for each statement. 

Statistics Views

Each release brings more of the database statistics out into a more accessible place, which is the database catalogs. Programmers can query these catalogs by the application to gain a rich variety of information, including statistics about table, index, and table partition usage in their database. In V6R1, several new views were added to the catalogs. Note that the first five views in the table have also been PTFed back to V5R4.

Application Development

Client special registers will be very useful. Application programmers can set their own text in these registers when their applications connect and use the database. By specifying this information, programmers can have each of their client applications tagged with their own "signature." The client register information can be captured by looking at the job or by collecting an SQL Performance Monitor. This should be helpful when analyzing performance, troubleshooting, or just gathering statistics about the different client applications.

These special registers can be changed using the following:

• The Set Client Information (SQLESETI) API
• In CLI, the SQLSetConnectAttr() API
• In JDBC, the setClientInfo connection method

There have been several CLI enhancements made for application development. These include support for wide APIs, alias support in Metadata APIs, row-wise array INSERT, and complete ISO timestamp support.

JDBC has been upgraded to Version 4.0, and .Net support has been enhanced with distributed transaction support and multi-row INSERT. In addition, .Net has better integration with Visual Studio.

One of the more widely used enhancements will be the added support in embedded SQL for RPG precompiler variable scoping for local variables at a procedure level. Also, all the ILE SQL precompilers have added stream file support so that a stream file can now contain SQL statements. In addition, Unicode support was added into the COBOL SQL precompiler.

For those that use the RUNSQLSTM command, support has been added to allow running SQL statements that are stored in a stream file. In addition, SQL statements in a source member are no longer limited to the first 80 characters of each line; a right margin parameter has been added to the command to specify where the end of the line should be. It has also been enhanced to allow CL commands to be run within RUNSQLSTM. These enhancements, coupled with the ability of System i Navigator's Run SQL Scripts to use source physical file members and IFS files will help simplify storing/sharing and the running of SQL statements in programming environments.

SQL Query Engine (SQE) and Runtime Performance Improvements

SQE enhancements include the use of National Language Sort Sequence (NLSS) during optimization, as well as the ability to handle more CASE statements, UDTFs, and lateral correlation, to name a few. There are many other new capabilities and performance enhancements as well.

Prior to V6R1, a QAQQINI parameter called Ignore_Derived_Index was introduced with a default of *NO. This was because queries that referenced DDS files would be sent down the CQE path and programmers would not want derived keys ignored. With V6R1, the default value of the Ignore_Derived_Index QAQQINI parameter is now *YES because programmers can now use the new CREATE INDEX syntax to define derived keys instead of DDS.

SQL runtime improvements have also been made to the database in V6R1. These include host server variable-length compression, SQL Call caching improvements, and MQT REFRESH improvements. Also, an improvement has been made to the RGZPFM command to eliminate the need for an exclusive lock when returning storage. ALTER TABLE now preserves indexes in tables with LOB columns, and performance has been improved by not requiring the rebuild of EVIs.

Database Limits Have Been Increased

SQL application and runtime limits have been increased. These include 128-byte cursor and statement names, full 64K result-set row-length support, ability to allow more than