Overview | Index Types | Index Options | Index Groups

Index Groups are used to combine multiple indexes into one single index. This allows a search against any one index within the group to automatically search all indexes within the group. For example, if a table has three lines, ADDRESS_LINE_1, ADDRESS_LINE_2, and ADDRESS_LINE_3, they can be indexed as an Index Group. Criteria against any one of those three columns will automatically search all address lines.

In the example below, a row displays an address:

ADDRESS_LINE_1          1254 Broadway Street
ADDRESS_LINE_2          Suite 100
ADDRESS_LINE_3          Yonkers, NY  10710

If these three columns are added to an Index Group, then the following search will locate the row:

  select        NAME, ADDRESS_LINE_1, ADDRESS_LINE_2, ADDRESS_LINE_3
    from        CUSTOMERS
    where       ADDRESS_LINE_1 = '1254 Broadway, Yonkers, NY'

Care should be taken when considering where to use Index Groups. When indexes are combined into an Index Group, they lose the ability to distinguish which column originally contained the values. This can lead to confusion. For example, the following search will also locate the row:

  select        NAME, ADDRESS_LINE_1, ADDRESS_LINE_2, ADDRESS_LINE_3
    from        CUSTOMERS
    where       ADDRESS_LINE_1 = '10710 Yonkers Street Suite 100'

Index Groups can be used in many different situations. Here is a common situation where Index Groups are used:

• Combining indexes for multiple address lines
• Combining indexes for a short-description column and a long-description column.
• Combining indexes for given names, nicknames, and aliases
• Combining indexes for multiple telephone numbers
• Combining indexes for multiple email addresses
• Combining indexes for multiple SIC or NAICS codes, or similar identifiers

Index Groups can combine indexes from multiple pre-joined tables. This can be particularly valuable when a search in a parent table needs to also search a child table at the same time. In this example, a HOUSEHOLDS table has a primary phone number and an INDIVIDUALS table has a mobile phone number for each member of the household.

Parent row:

HOUSEHOLDS.HOUSEHOLD      1234
HOUSEHOLDS.HOME_PHONE     303 449 2198

Child row:

INDIVIDUALS.HOUSEHOLD     1234
INDIVIDUALS.INDIVIDUAL    5678  
INDIVIDUALS.MOBILE_PHONE  720 298 3162

Child row:

INDIVIDUALS.HOUSEHOLD     1234
INDIVIDUALS.INDIVIDUAL    5679  
INDIVIDUALS.MOBILE_PHONE  720 675 2238

If HOUSEHOLDS.HOME_PHONE and INDIVIDUALS.MOBILE_PHONE were added to an Index Group, then this household would be returned in the following query. Note that the query searches the indexes in the INDIVIDUALS table even though the INDIVIDUALS table is not referenced in the query.

  select        HOUSEHOLD, ADDRESS, CITY, STATE, ZIP
    from        HOUSEHOLDS
    where       HOME_PHONE = '720 298 3162'

As discussed above, care should be taken when considering where to use Index Groups across tables. It can be confusing when a table is actually searched without it being referenced in the SQL statement. A row can also be qualified when a portion of the criteria exists in the parent and another portion exists in a child row. For example, the following search will also locate this household:

  select        HOUSEHOLD, ADDRESS, CITY, STATE, ZIP
    from        HOUSEHOLDS
    where       HOME_PHONE = '303 298 2238'

Because of this characteristic, Index Groups are best suited for a) Omnidex Indexes, b) QuickText Indexes containing free-form text where the location of keywords is not material, or c) combinations of indexes where the data between columns is so different it would not be confused.

Prev

See also: