Apache Spark Performance - AWS

I am a newbie to apache spark. My setup has 3 nodes (AWS M4 large) each with hadoop and apache spark installed. I have a file with 10Million records in comma separated file (approximately 1.5GB). This file is copied on HDFS and with replication factor of 2. I have a spark scala program (built into jar). Program reads the HDFS file, splits the records by commas, makes a keyvalue pair of one of the columns and count of that column. Finally, i sort the keyvalue pairs (K,V) by the value field (V) to get highest counts first. Then i take 5 records and print them.

Program is to get top 5 talkers in the file.

I get the results in 35 seconds. I see spark distributing the work and i see best data locality also. Just wondering if 35 seconds is expected for this kind of operation on apache spark. Thanks in advance for time spent reading this message.