Guide to Developing and Deploying Custom Partition Algorithms in DBLE

DBLE is a distributed middleware that supports many built‑in data‑splitting algorithms; it also allows developers to create custom partition algorithms to meet specific business needs.

1. How functions are loaded – When DBLE starts or reloads, it reads rule.xml, uses the class attribute to locate the corresponding JAR via Java reflection, creates an instance, and invokes setter methods to assign configuration values.

2. Routing calculation – The routing function receives the sharding key value from the user SQL, computes the logical partition number, and directs the query to the appropriate data slice.

3. Parameter query – DBLE calls getAllProperties() to expose the configuration map defined by the setters.

4. Development and deployment – Developers should reference the whole DBLE source or include it as a project dependency, implement AbstractPartitionAlgorithm and RuleAlgorithm, package the implementation as a JAR, place it in $DBLE_HOME/lib (or $DBLE_HOME/algorithm for newer versions), adjust permissions, and reference the fully‑qualified class name in rule.xml.

5. Interface specification – A custom routing class must provide configuration setters (e.g., setPartitionCount(String), setPartitionLength(String)), a selfCheck() method for validation, an init() method for initialization, calculate() / calculateRange() for single‑value and range queries, and getAllProperties() to return the configuration map.

6. Example code snippets – The article includes typical setter implementations, a sample selfCheck() stub, and full implementations of calculate() and calculateRange() showing error handling and partition range logic.

7. Built‑in function abbreviations – A table lists the short names (e.g., date, hash, stringhash) and their corresponding Java classes (e.g., PartitionByDate, PartitionByLong).