public class HTreeVersusHashMapDemo extends Object
HTreeperformance against a Java collections class such as
LinkedHashMap. This demonstration is focused on the performance curve when inserting a large number of keys into a collection. The Java collection classes are faster for small numbers of keys, but the
HTreerapidly out performs them as the #of keys grows larger.
HTree is run against the
MemoryManager. This means that
the data for the
HTree is mostly stored on the Java native process
ByteBuffer.allocateDirect(int). Thus even a very
HTree instance can be run with a very small JVM object heap and
creates nearly no GC pressure.
HTree permits multiple entries for the same key. While variable
length keys are supported, the key for the
HTree is typically a 32
(or 64) bit hash code for the value stored in the
the application will lookup all tuples having the same hash code and then
Map semantics by scanning those tuples for the
presence of the same application object.
|Constructor and Description|
public static void main(String args)
Copyright © 2006–2016 SYSTAP, LLC DBA Blazegraph. All rights reserved.