Replace two maps and a union with a flatMap. Seems to be faster

pythonApp/netflowAlgs.py

@@ -31,7 +31,9 @@ def ports_count_by_ip3(csv):
 
 
 def ports_count_by_ip(csv):
-    srcs = csv.map(lambda x: ((x[DEST_PORT], x[SRC_IP]), 1))
-    dsts = csv.map(lambda x: ((x[DEST_PORT], x[DEST_IP]), 1))
-    ips = srcs.union(dsts).reduceByKey(add)
+#    srcs = csv.map(lambda x: ((x[DEST_PORT], x[SRC_IP]), 1))
+#    dsts = csv.map(lambda x: ((x[DEST_PORT], x[DEST_IP]), 1))
+#    ips = srcs.union(dsts).reduceByKey(add)
+    ips = csv.flatMap(lambda x: (((x[DEST_PORT], x[DEST_IP]), 1),
+                                 ((x[DEST_PORT], x[SRC_IP]), 1))).reduceByKey(add)
     return ips.map(lambda x: (x[1], x[0])).sortByKey(False).take(20)