Defining Data Tables

This section explains how to configure data tables to store the events you need to persist to carry out time series aggregation.

The data handled by WSO2 Stream Processor are stored in the following two types of tables:

In-memory tables: If no store-backed tables are defined, data is stored in in-memory tables by default.
Store-backed tables: These are tables that are defined by you in an external database. For a list of database types supported and instructions to define table for different database types, see Defining Tables for External Data Stores.

Adding primary and index keys

Both in-memory tables and tables backed by external databases support primary and index keys. These are defined to allow stored information to be searched and retrieved in an effective and efficient manner.

Adding primary keys

Attribute(s) within the event stream for which the event table is created can be specified as the primary key for the table. The purpose of primary key is to ensure that the value for a selected attribute is unique for each entry in the table. This prevents the duplication of entries saved in the table.

Primary keys are configured via the @PrimaryKey annotation. Only one @PrimaryKey annotation is allowed per event table.

When several attributes are given within Primary key annotation (e.g @PrimaryKey( 'key1', 'key2')), those attributes would act as a composite primary key.

Syntax

@PrimaryKey('<attribute_1>')
define table <event_table> (<attribute_1> <attribute_type>, <attribute_2> <attribute_type>, <attribute_3> <attribute_type>);

Example

@PrimaryKey('symbol')
define table StockTable (symbol string, price float, volume long);

The above configuration ensures that each entry saved in the StockTable event table should have a unique value for the symbol attribute because this attribute is defined as the primary key.

Adding indexes

An attribute within the event stream for which the event table is created can be specified as the primary key for the table. This allows the entries stored within the table to be indexed by that attribute.

Indexes are configured via the @Index annotation.

An event table can have multiple attributes defined as index attributes. However, only one @Index annotation can be added per event table.

Syntax

To index by a single attribute:

@Index('<attribute_1>')
define table <event_table> (<attribute_1> <attribute_type>, <attribute_2> <attribute_type>, <attribute_3> <attribute_type>);

To index by multiple attributes:

@Index('<attribute_1>''<attribute_2>')
define table <event_table> (<attribute_1> <attribute_type>, <attribute_2> <attribute_type>, <attribute_3> <attribute_type>);

Example

@Index('symbol')
define table StockTable (symbol string, price float, volume long);

The above configuration ensures that the entries stored in the StockTable event table are indexed by the symbol attribute.