You can call the ComputeSplitsBySize operation to logically splits the data in a table into multiple shards whose sizes are close to a specified value. The locations of the split points among these splits and the hosts where the shards reside are returned. This operation is typically used to implement plans on compute engines, such as concurrency plans.

Note For more information about ComputeSplitsBySize, see ComputeSplitPointsBySize.

Prerequisites

  • OTSClient is initialized. For more information, see Initialization.
  • A data table is created, and data is written to the table.

Parameters

Parameter Description
tableName The name of the base table.
splitSize The approximate size of each shard.

Unit: 100 MB.

Examples

The following code provides an example on how to logically splits the data in a table into multiple shards whose sizes are close to 200 MB: 

private static void describeTable(SyncClient client) {
    // Splits the data in the table into multiple shards based on the size of 200 MB.
    ComputeSplitsBySizeRequest request = new ComputeSplitsBySizeRequest(TABLE_NAME, 2);
    ComputeSplitsBySizeResponse response = client.computeSplitsBySize(request);
    System.out.println("ConsumedCapacity="   response.getConsumedCapacity().jsonize());
    System.out.println("PrimaryKeySchema="   response.getPrimaryKeySchema());
    System.out.println("RequestId="   response.getRequestId());
    System.out.println("TraceId="   response.getTraceId());
    List<Split> splits = response.getSplits();
    System.out.println("splits.size="   splits.size());
    Iterator<Split> iterator = splits.iterator();
    while (iterator.hasNext()) {
        Split split = iterator.next();
        System.out.println("split.getLocation()="   split.getLocation());
        // You can inject split.getLowerBound() and split.getUpperBound() into RangeRowQueryCriteria, and pass RangeRowQueryCriteria to getRange() or createRangeIterator().
        System.out.println("split.getLowerBound()="   split.getLowerBound().jsonize());
        System.out.println("split.getUpperBound()="   split.getUpperBound().jsonize());
    }
}