/// <summary>
/// Base size of blocks of entries. As entries gets written to a BlockStorage, they are buffered up to this size, then sorted and written out.
/// As blocks gets merged into bigger blocks, this is still the size of the read buffer for each block no matter its size.
/// Each thread has its own buffer when writing and each thread has <seealso cref="mergeFactor"/> buffers when merging.
/// The memory usage will be at its biggest during merge and a total memory usage sum can be calculated like so:
///
/// blockSize * numberOfPopulationWorkers * <seealso cref="mergeFactor"/>
///
/// where typically <seealso cref="BatchingMultipleIndexPopulator"/> controls the number of population workers. The setting
/// `unsupported.dbms.multi_threaded_schema_index_population_enabled` controls whether or not the multi-threaded <seealso cref="BatchingMultipleIndexPopulator"/>
/// is used, otherwise a single-threaded populator is used instead.
/// </summary>
public static int ParseBlockSize()
{
long blockSize = ByteUnit.parse(FeatureToggles.getString(typeof(BlockBasedIndexPopulator), BLOCK_SIZE_NAME, "1M"));
Preconditions.checkArgument(blockSize >= 20 && blockSize < int.MaxValue, "Block size need to fit in int. Was " + blockSize);
return(( int )blockSize);
}
