From an architectural perspective, you can transform your data in the following ways:

• Multistage Data Transformation in Data Warehouses

• Pipelined Data Transformation in Data Warehouses

• Staging Area in Data Warehouses

OCI and direct-path APIs are frequently used when the transformation and computation are done outside the database and there is no need for flat file staging.

Export and import are used when the data is inserted as is into the target system. No complex extractions are possible. See Extraction in Data Warehouses for further information.

In a data warehouse environment, you can use procedural languages such as PL/SQL to implement complex transformations in the Oracle Database. Whereas CTAS operates on entire tables and emphasizes parallelism, PL/SQL provides a row-based approached and can accommodate very sophisticated transformation rules. For example, a PL/SQL procedure could open multiple cursors and read data from multiple source tables, combine this data using complex business rules, and finally insert the transformed data into one or more target table. It would be difficult or impossible to express the same sequence of operations using standard SQL statements.

Using a procedural language, a specific transformation (or number of transformation steps) within a complex ETL processing can be encapsulated, reading data from an intermediate staging area and generating a new table object as output. A previously generated transformation input table and a subsequent transformation will consume the table generated by this specific transformation. Alternatively, these encapsulated transformation steps within the complete ETL process can be integrated seamlessly, thus streaming sets of rows between each other without the necessity of intermediate staging. You can use table functions to implement such behavior.

Table functions provide the support for pipelined and parallel execution of transformations implemented in PL/SQL, C, or Java. Scenarios as mentioned earlier can be done without requiring the use of intermediate staging tables, which interrupt the data flow through various transformations steps. Detailed information about table functions is provided in "What is a Table Function?".

Data that is logically not clean violates business rules that are known prior to any data consumption. Most of the time, handling these kind of errors will be incorporated into the loading or transformation process. However, in situations where the error identification for all records would become too expensive and the business rule can be enforced as a data rule violation, for example, testing hundreds of columns to see if they are NOT NULL, programmers often choose to handle even known possible logical error cases more generically. An example of this is shown in "Data Error Scenarios".

Incorporating logical rules can be as easy as applying filter conditions on the data input stream or as complex as feeding the dirty data into a different transformation workflow. Some examples are as follows:

• Filtering of logical data errors using SQL. Data that does not adhere to certain conditions is filtered out prior to being processed.

• Identifying and separating logical data errors. In simple cases, this can be accomplished using SQL, as shown in Example 19-1, or in more complex cases in a procedural approach, as shown in Example 19-6.

Unlike logical errors, data rule violations are not usually anticipated by the load or transformation process. Such unexpected data rule violations (also known as data errors) that are not handled from an operation cause the operation to fail. Data rule violations are error conditions that happen inside the database and cause a statement to fail. Examples of this are data type conversion errors or constraint violations.

In the past, SQL did not offer a way to handle data errors on a row level as part of its bulk processing. The only way to handle data errors inside the database was to use PL/SQL. Now, however, you can log data errors into a special error table while the DML operation continues. You can also handle data conversion errors using SQL functions.

The following sections briefly discuss the various exception handling strategies:

• Handling Data Errors with SQL

• Handling Data Errors in PL/SQL

• Handling Data Errors with an Error Logging Table

A typical transformation is the key lookup. For example, suppose that sales transaction data has been loaded into a retail data warehouse. Although the data warehouse's sales table contains a product_id column, the sales transaction data extracted from the source system contains Uniform Price Codes (UPC) instead of product IDs. Therefore, it is necessary to transform the UPC codes into product IDs before the new sales transaction data can be inserted into the sales table.

In order to execute this transformation, a lookup table must relate the product_id values to the UPC codes. This table might be the product dimension table, or perhaps another table in the data warehouse that has been created specifically to support this transformation. For this example, you assume that there is a table named product, which has a product_id and an upc_code column.

This data substitution transformation can be implemented using the following CTAS statement:

CREATE TABLE temp_sales_step2 NOLOGGING PARALLEL AS SELECT sales_transaction_id,
  product.product_id sales_product_id, sales_customer_id, sales_time_id,
  sales_channel_id, sales_quantity_sold, sales_dollar_amount
FROM  temp_sales_step1, product
WHERE temp_sales_step1.upc_code = product.upc_code;

This CTAS statement converts each valid UPC code to a valid product_id value. If the ETL process has guaranteed that each UPC code is valid, then this statement alone may be sufficient to implement the entire transformation.

In the preceding example, if you must also handle new sales data that does not have valid UPC codes (a logical data error), you can use an additional CTAS statement to identify the invalid rows:

CREATE TABLE temp_sales_step1_invalid NOLOGGING PARALLEL AS
SELECT * FROM temp_sales_step1 s
WHERE NOT EXISTS (SELECT 1 FROM product p WHERE p.upc_code=s.upc_code);

This invalid data is now stored in a separate table, temp_sales_step1_invalid, and can be handled separately by the ETL process.

Another way to handle invalid data is to modify the original CTAS to use an outer join, as in the following statement:

CREATE TABLE temp_sales_step2 NOLOGGING PARALLEL AS
SELECT sales_transaction_id, product.product_id sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id, sales_quantity_sold,
   sales_dollar_amount
FROM  temp_sales_step1, product
WHERE temp_sales_step1.upc_code = product.upc_code (+);

Using this outer join, the sales transactions that originally contained invalidated UPC codes are assigned a product_id of NULL. These transactions can be handled later. Alternatively, you could use a multi-table insert, separating the values with a product_id of NULL into a separate table; this might be a beneficial approach when the expected error count is relatively small compared to the total data volume. You do not have to touch the large target table but only a small one for a subsequent processing.

INSERT /*+ APPEND PARALLEL */ FIRST
WHEN sales_product_id IS NOT NULL THEN
   INTO temp_sales_step2
   VALUES (sales_transaction_id, sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount)
ELSE
   INTO temp_sales_step1_invalid
   VALUES (sales_transaction_id, sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount)
SELECT sales_transaction_id, product.product_id sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount
FROM temp_sales_step1, product
WHERE temp_sales_step1.upc_code = product.upc_code (+);

Note that for this solution, the empty tables temp_sales_step2 and temp_sales_step1_invalid must already exist.

Additional approaches to handling invalid UPC codes exist. Some data warehouses may choose to insert null-valued product_id values into their sales table, while others may not allow any new data from the entire batch to be inserted into the sales table until all invalid UPC codes have been addressed. The correct approach is determined by the business requirements of the data warehouse. Irrespective of the specific requirements, exception handling can be addressed by the same basic SQL techniques as transformations.

If the quality of the data is unknown, the example discussed in Business Rule Violation Scenario could be enhanced to handle unexpected data errors, for example, data type conversion errors, as shown in the following:

INSERT /*+ APPEND PARALLEL */ FIRST
WHEN sales_product_id IS NOT NULL THEN
INTO temp_sales_step2
VALUES (sales_transaction_id, sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount)
LOG ERRORS INTO sales_step2_errors('load_20040804')
REJECT LIMIT UNLIMITED
ELSE
INTO temp_sales_step1_invalid
VALUES (sales_transaction_id, sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount)
LOG ERRORS INTO sales_step2_errors('load_20040804')
REJECT LIMIT UNLIMITED
SELECT sales_transaction_id, product.product_id sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount
FROM temp_sales_step1, product
WHERE temp_sales_step1.upc_code = product.upc_code (+);

This statement tracks the logical data error of not having a valid product UPC code in table temp_sales_step1_invalid and all other possible errors in a DML error logging table called sales_step2_errors. Note that an error logging table can be used for several DML operations.

An alternative to this approach would be to enforce the business rule of having a valid UPC code on the database level with a NOT NULL constraint. Using an outer join, all orders not having a valid UPC code would be mapped to a NULL value and then treated as data errors. This DML error logging capability is used to track these errors in the following statement:

INSERT /*+ APPEND PARALLEL */
INTO temp_sales_step2
VALUES (sales_transaction_id, sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount)
SELECT sales_transaction_id, product.product_id sales_product_id,
   sales_customer_id, sales_time_id, sales_channel_id,
   sales_quantity_sold, sales_dollar_amount
FROM temp_sales_step1, product
WHERE temp_sales_step1.upc_code = product.upc_code (+)
LOG ERRORS INTO sales_step2_errors('load_20040804')
REJECT LIMIT UNLIMITED;

The error logging table contains all records that would have caused the DML operation to fail. You can use its content to analyze and correct any error. The content in the error logging table is preserved for any DML operation, irrespective of the success of the DML operation itself. Let us assume the following SQL statement failed because the reject limit was reached:

SQL> INSERT /*+ APPEND NOLOGGING PARALLEL */ INTO sales_overall
2 SELECT * FROM sales_activity_direct
3 LOG ERRORS INTO err$_sales_overall ('load_test2')
4 REJECT LIMIT 10;
SELECT * FROM sales_activity_direct
*
ERROR at line 2:
ORA-01722: invalid number

The name of the error logging table, err$_sales_overall, is the default derived by using the DBMS_ERRLOG package. See Oracle Database PL/SQL Packages and Types Reference for more information.

The error message raised by Oracle occurs where the first after the error limit is reached. The next error (number 11) is the one that raised an error. The error message that is displayed is based on the error that exceeded the limit, so, for example, the ninth error could be different from the eleventh error.

The target table sales_overall will not show any records being entered (assumed that the table was empty before), but the error logging table will contain 11 rows (REJECT LIMIT + 1)

SQL> SELECT COUNT(*) FROM sales_overall;
COUNT(*)
----------
0

SQL> SELECT COUNT(*) FROM err$_sales_overall;
COUNT(*)
----------
11

A DML error logging table consists of several fixed control columns that are mandatory for every error logging table. Besides the Oracle error number, Oracle enforces storing the error message as well. In many cases, the error message provides additional information to analyze and resolve the root cause for the data error. The following SQL output of a DML error logging table shows this difference. Note that the second output contains the additional information for rows that were rejected due to NOT NULL violations.

SQL> SELECT DISTINCT ora_err_number$ FROM err$_sales_overall;

ORA_ERR_NUMBER$
---------------
           1400
           1722
           1830
           1847

SQL> SELECT DISTINCT ora_err_number$, ora_err_mesg$ FROM err$_sales_overall;

ORA_ERR_NUMBER$       ORA_ERR_MESG$
           1400       ORA-01400: cannot insert NULL into
                      ("SH"."SALES_OVERALL"."CUST_ID")
           1400       ORA-01400: cannot insert NULL into
                      ("SH"."SALES_OVERALL"."PROD_ID")
           1722       ORA-01722: invalid number
           1830       ORA-01830: date format picture ends before
                      converting entire input string
           1847       ORA-01847: day of month must be between 1 and last
                      day of month

A data warehouse can receive data from many different sources. Some of these source systems may not be relational databases and may store data in very different formats from the data warehouse. For example, suppose that you receive a set of sales records from a nonrelational database having the form:

product_id, customer_id, weekly_start_date, sales_sun, sales_mon, sales_tue,
  sales_wed, sales_thu, sales_fri, sales_sat

The input table looks like the following:

SELECT * FROM sales_input_table;

PRODUCT_ID CUSTOMER_ID WEEKLY_ST  SALES_SUN  SALES_MON  SALES_TUE  SALES_WED SALES_THU  SALES_FRI  SALES_SAT
---------- ----------- --------- ---------- ---------- ---------- -------------------- ---------- ----------
       111         222 01-OCT-00        100        200        300        400       500        600        700
       222         333 08-OCT-00        200        300        400        500       600        700        800
       333         444 15-OCT-00        300        400        500        600       700        800        900

In your data warehouse, you would want to store the records in a more typical relational form in a fact table sales of the sh sample schema:

prod_id, cust_id, time_id, amount_sold

Thus, you need to build a transformation such that each record in the input stream must be converted into seven records for the data warehouse's sales table. This operation is commonly referred to as pivoting, and Oracle Database offers several ways to do this.

The result of the previous example will resemble the following:

SELECT prod_id, cust_id, time_id, amount_sold FROM sales;

   PROD_ID    CUST_ID   TIME_ID   AMOUNT_SOLD
---------- ----------   --------- -----------
       111        222   01-OCT-00         100
       111        222   02-OCT-00         200
       111        222   03-OCT-00         300
       111        222   04-OCT-00         400
       111        222   05-OCT-00         500
       111        222   06-OCT-00         600
       111        222   07-OCT-00         700
       222        333   08-OCT-00         200
       222        333   09-OCT-00         300
       222        333   10-OCT-00         400
       222        333   11-OCT-00         500
       222        333   12-OCT-00         600
       222        333   13-OCT-00         700
       222        333   14-OCT-00         800
       333        444   15-OCT-00         300
       333        444   16-OCT-00         400
       333        444   17-OCT-00         500
       333        444   18-OCT-00         600
       333        444   19-OCT-00         700
       333        444   20-OCT-00         800
       333        444   21-OCT-00         900

INSERT ALL INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date, sales_sun)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+1, sales_mon)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+2, sales_tue)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+3, sales_wed)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+4, sales_thu)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+5, sales_fri)
      INTO sales (prod_id, cust_id, time_id, amount_sold)
      VALUES (product_id, customer_id, weekly_start_date+6, sales_sat)
SELECT product_id, customer_id, weekly_start_date, sales_sun,
      sales_mon, sales_tue, sales_wed, sales_thu, sales_fri, sales_sat
FROM sales_input_table;