com.facebook.presto.spi.ConnectorTablePartitioning Maven / Gradle / Ivy
/*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.facebook.presto.spi;
import com.facebook.presto.spi.connector.ConnectorNodePartitioningProvider;
import com.facebook.presto.spi.connector.ConnectorPartitioningHandle;
import com.facebook.presto.spi.connector.ConnectorSplitManager;
import java.util.ArrayList;
import java.util.List;
import java.util.Objects;
import static java.util.Collections.unmodifiableList;
import static java.util.Objects.requireNonNull;
/**
* The partitioning of split groups in a table.
* A table is partitioned on columns {C1,...,Cn} if rows with the same values
* on the respective columns are always in the same split group.
*
* This class was previously name ConnectorNodePartitioning.
* The new semantics is more flexible.
* It enables expression of more sophisticated table organization.
* Node partitioning is a form of partitioning between split groups,
* where each split group contains all splits on a {@link Node}.
*
* Unless the connector declares itself as supporting grouped scheduling in
* {@link ConnectorSplitManager}, Presto engine treats all splits on a node
* as a single split group.
* As a result, the connector does not need to provide additional guarantee
* as a result of this change for previously-declared node partitioning.
* Therefore, this change in SPI is backward compatible.
*
* For now, all splits in each split group must be assigned the same {@link Node}
* by {@link ConnectorNodePartitioningProvider}.
* With future changes to the engine, connectors will no longer be required
* to declare a mapping from split groups to nodes.
* Artificially requiring such a mapping regardless of whether the engine can
* take advantage of the TablePartitioning negatively affects performance.
*/
public class ConnectorTablePartitioning
{
private final ConnectorPartitioningHandle partitioningHandle;
private final List partitioningColumns;
public ConnectorTablePartitioning(ConnectorPartitioningHandle partitioningHandle, List partitioningColumns)
{
this.partitioningHandle = requireNonNull(partitioningHandle, "partitioningHandle is null");
this.partitioningColumns = unmodifiableList(new ArrayList<>(requireNonNull(partitioningColumns, "partitioningColumns is null")));
}
/**
* A handle to the partitioning scheme used to divide the table across worker nodes.
*/
public ConnectorPartitioningHandle getPartitioningHandle()
{
return partitioningHandle;
}
/**
* The columns used to partition the table across the worker nodes. An empty list means
* the table is entirely contained on a single worker.
*
* If the table is node partitioned, the connector guarantees that each combination of values for
* the distributed columns will be contained within a single worker.
*/
public List getPartitioningColumns()
{
return partitioningColumns;
}
@Override
public boolean equals(Object o)
{
if (this == o) {
return true;
}
if (o == null || getClass() != o.getClass()) {
return false;
}
ConnectorTablePartitioning that = (ConnectorTablePartitioning) o;
return Objects.equals(partitioningHandle, that.partitioningHandle) &&
Objects.equals(partitioningColumns, that.partitioningColumns);
}
@Override
public int hashCode()
{
return Objects.hash(partitioningHandle, partitioningColumns);
}
}