I want to use a Stream to parallelize processing of a heterogenous set of remotely stored JSON files of unknown number (the number of files is not known upfront). The files can vary widely in size, from 1 JSON record per file up to 100,000 records in some other files. A JSON record in this case means a self-contained JSON object represented as one line in the file.
I really want to use Streams for this and so I implemented this Spliterator:
public abstract class JsonStreamSpliterator<METADATA, RECORD> extends AbstractSpliterator<RECORD> {
abstract protected JsonStreamSupport<METADATA> openInputStream(String path);
abstract protected RECORD parse(METADATA metadata, Map<String, Object> json);
private static final int ADDITIONAL_CHARACTERISTICS = Spliterator.IMMUTABLE | Spliterator.DISTINCT | Spliterator.NONNULL;
private static final int MAX_BUFFER = 100;
private final Iterator<String> paths;
private JsonStreamSupport<METADATA> reader = null;
public JsonStreamSpliterator(Iterator<String> paths) {
this(Long.MAX_VALUE, ADDITIONAL_CHARACTERISTICS, paths);
}
private JsonStreamSpliterator(long est, int additionalCharacteristics, Iterator<String> paths) {
super(est, additionalCharacteristics);
this.paths = paths;
}
private JsonStreamSpliterator(long est, int additionalCharacteristics, Iterator<String> paths, String nextPath) {
this(est, additionalCharacteristics, paths);
open(nextPath);
}
@Override
public boolean tryAdvance(Consumer<? super RECORD> action) {
if(reader == null) {
String path = takeNextPath();
if(path != null) {
open(path);
}
else {
return false;
}
}
Map<String, Object> json = reader.readJsonLine();
if(json != null) {
RECORD item = parse(reader.getMetadata(), json);
action.accept(item);
return true;
}
else {
reader.close();
reader = null;
return tryAdvance(action);
}
}
private void open(String path) {
reader = openInputStream(path);
}
private String takeNextPath() {
synchronized(paths) {
if(paths.hasNext()) {
return paths.next();
}
}
return null;
}
@Override
public Spliterator<RECORD> trySplit() {
String nextPath = takeNextPath();
if(nextPath != null) {
return new JsonStreamSpliterator<METADATA,RECORD>(Long.MAX_VALUE, ADDITIONAL_CHARACTERISTICS, paths, nextPath) {
@Override
protected JsonStreamSupport<METADATA> openInputStream(String path) {
return JsonStreamSpliterator.this.openInputStream(path);
}
@Override
protected RECORD parse(METADATA metaData, Map<String,Object> json) {
return JsonStreamSpliterator.this.parse(metaData, json);
}
};
}
else {
List<RECORD> records = new ArrayList<RECORD>();
while(tryAdvance(records::add) && records.size() < MAX_BUFFER) {
// loop
}
if(records.size() != 0) {
return records.spliterator();
}
else {
return null;
}
}
}
}
The problem I'm having is that while the Stream parallelizes beautifully at first, eventually the largest file is left processing in a single thread. I believe the proximal cause is well documented: the spliterator is "unbalanced".
More concretely, appears that the trySplit method is not called after a certain point in the Stream.forEach's lifecycle, so the extra logic to distribute small batches at the end of trySplit is rarely executed.
Notice how all the spliterators returned from trySplit share the same paths iterator. I thought this was a really clever way to balance the work across all spliterators, but it hasn't been enough to achieve full parallelism.
I would like the parallel processing to proceed first across files, and then when few large files are still left spliterating, I want to parallelize across chunks of the remaining files. That was the intent of the else block at the end of trySplit.
Is there an easy / simple / canonical way around this problem?
