Package com.linkedin.venice.partitioner
Class VenicePartitioner
java.lang.Object
com.linkedin.venice.partitioner.VenicePartitioner
- Direct Known Subclasses:
ConstantVenicePartitioner
,DefaultVenicePartitioner
,InvalidKeySchemaPartitioner
,SimplePartitioner
Determines partitioning, which is used for producing messages into the right
Kafka partitions and routing reads to the correct Venice storage nodes.
N.B.: This is purposefully de-coupled from Kafka, so that the Router does
not need to depend on Kafka.
-
Field Summary
-
Constructor Summary
-
Method Summary
Modifier and TypeMethodDescriptionprotected void
checkSchema
(org.apache.avro.Schema keySchema) Implementors of this class can optionally provide an implementation of this function, which can perform validation of schemas to be certain that they are compatible with the partitioner implementation.abstract int
getPartitionId
(byte[] keyBytes, int numPartitions) A function that returns the partitionId based on the key and partition count.int
getPartitionId
(byte[] keyBytes, int offset, int length, int numPartitions) Implementors of this class can optionally provide an implementation of this function, which would result in eliminating an instantiation ofByteBuffer
in the case where the provided offset and length do not map to the boundaries of the byte[].abstract int
getPartitionId
(ByteBuffer keyByteBuffer, int numPartitions)
-
Field Details
-
props
-
-
Constructor Details
-
VenicePartitioner
public VenicePartitioner() -
VenicePartitioner
-
VenicePartitioner
-
-
Method Details
-
getPartitionId
public abstract int getPartitionId(byte[] keyBytes, int numPartitions) A function that returns the partitionId based on the key and partition count.- Parameters:
keyBytes
- - A key that will be mapped into a partitionnumPartitions
- - The number of total partitions available in Kafka/storage- Returns:
-
getPartitionId
public int getPartitionId(byte[] keyBytes, int offset, int length, int numPartitions) Implementors of this class can optionally provide an implementation of this function, which would result in eliminating an instantiation ofByteBuffer
in the case where the provided offset and length do not map to the boundaries of the byte[]. This is just a minor optimization. -
getPartitionId
-
checkSchema
protected void checkSchema(@Nonnull org.apache.avro.Schema keySchema) throws PartitionerSchemaMismatchException Implementors of this class can optionally provide an implementation of this function, which can perform validation of schemas to be certain that they are compatible with the partitioner implementation.- Parameters:
keySchema
- the schema to be validated- Throws:
PartitionerSchemaMismatchException
- should the provided schema not match the partitioner (with a message that explains why).
-