Overview

Database file partitioning provides:

• Easier management of large amounts of historical data, particularly for bulk data operations such as backups and deletions

• Cost savings through the use of tiered disk storage, which enables you to split database files into partitions, with different partitions then being able to be stored on different media devices, tiered disk storage, or both different media devices and tiered disk storage

• The ability to move objects between partitions programmatically, which enables you to avoid accessing historical information if it is not required, thereby improving performance of online and batch operations

A logical database file can be split into several parts, or partitions, where each partition corresponds to a physical file at the file system level. A logical database file is identified by its schema-defined name and is addressable by its name or number. A file partition is addressable by a DbFile instance number and a unique partition identifier (JadeDbFilePartition class partitionID property), which is a positive Integer64 value in the range 1 through 2^32‑15. This unique identifier remains associated with a partition for its lifetime.

When the setPartitioned method is executed for an instance of the DbFile class representing a new database file, the required partition structures are established when the file is first instantiated. Making an existing database file partitioned requires a reorganization to convert the standard map file to the partitioned structure.

Partitioning a class enables you to locate instances in different partitions of a database file. A file partition is therefore a part of a database file containing a subset of instances of a single class.

When a database file has been partitioned, the set of partitions associated with a schema-defined map file is referred to as a partition set.

The external file name assigned to a file partition is derived from the schema-defined file name and incorporating the partition number as a qualifier, with the following format.

<map_file_name>_part<partition-number>.dat

For details about replicating partitioned file structures and most meta data on SDS secondary databases, see "SDS and Recovery Considerations when Using Partitioned Database Files", in Chapter 1 of the JADE Synchronized Database Service (SDS) Administration Guide.

The JadeDbFilePartition class provides methods that enable you to freeze or thaw a file partition. When a partition is frozen:

• The partition is converted to a read-only state, allowing its relocation to read-only media

• Objects read from the partition are marked as frozen, overriding individual object volatility state

When a partition is thawed:

• The read-only status is removed, allowing updates to objects residing in the partition

• Volatility reverts to transparent, removing the frozen override