Added Scale Factor graphalytics.3000

Added possibility of setting distribution parameters from params.ini

Author: ArnauPrat

src/main/java/ldbc/snb/datagen/generator/LDBCDatagen.java

@@ -103,6 +103,13 @@ public int runGenerateJob(Configuration conf) throws Exception {
         fs.delete(new Path(personsFileName1), true);
         long endRandom= System.currentTimeMillis();
 
+/*        printProgress("Creating edges to fill the degree gap");
+        long startGap = System.currentTimeMillis();
+        knowsGenerator = new HadoopKnowsGenerator(conf,null, "ldbc.snb.datagen.hadoop.DegreeGapKeySetter", 1.0f);
+        knowsGenerator.run(personsFileName2,personsFileName1);
+        fs.delete(new Path(personsFileName2), true);
+        long endGap = System.currentTimeMillis();*/
+
         printProgress("Serializing persons");
         long startPersonSerializing= System.currentTimeMillis();
         HadoopPersonSerializer serializer = new HadoopPersonSerializer(conf);

src/main/java/ldbc/snb/datagen/generator/distribution/MOEZipfDistribution.java

@@ -26,6 +26,7 @@ public void initialize(Configuration conf) {
     }
 
     public void reset (long seed) {
+        random_.setSeed(seed);
         zipf_.reseedRandomGenerator(seed);
     }
 

src/main/java/ldbc/snb/datagen/hadoop/DegreeGapKeySetter.java

@@ -0,0 +1,14 @@
+package ldbc.snb.datagen.hadoop;
+
+import ldbc.snb.datagen.objects.Person;
+
+/**
+ * Created by aprat on 11/17/14.
+ */
+public class DegreeGapKeySetter implements HadoopFileKeyChanger.KeySetter<TupleKey> {
+
+    public TupleKey getKey(Object object ) {
+        Person person = (Person)object;
+        return new TupleKey(person.maxNumKnows()-person.knows().size(), person.accountId());
+    }
+}

src/main/resources/scale_factors.xml

@@ -267,4 +267,31 @@
             <value>false</value>
         </property>
     </scale_factor>
+
+    <scale_factor name="graphalytics.3000" >
+        <property>
+            <name>ldbc.snb.datagen.generator.numPersons</name>
+            <value>32500000</value>
+        </property>
+        <property>
+            <name>ldbc.snb.datagen.generator.startYear</name>
+            <value>2010</value>
+        </property>
+        <property>
+            <name>ldbc.snb.datagen.generator.numYears</name>
+            <value>1</value>
+        </property>
+        <property>
+            <name>ldbc.snb.datagen.generator.distribution.degreeDistribution</name>
+            <value>ldbc.snb.datagen.generator.distribution.FacebookDegreeDistribution</value>
+        </property>
+        <property>
+            <name>ldbc.snb.datagen.generator.activity</name>
+            <value>false</value>
+        </property>
+        <property>
+            <name>ldbc.snb.datagen.serializer.updateStreams</name>
+            <value>false</value>
+        </property>
+    </scale_factor>
 </scale_factors>