- Get your cluster ready with HW/SW
- Setup Passwordless ssh for pssh to work.
- Edit "userSettings.conf" and insert your intended variables. export BIG_BENCH_DEFAULT_DATABASE="bigbench" */ Name of Hive Metastore */ export BIG_BENCH_DEFAULT_ENGINE="hive" */ Default framework Engine */ export BIG_BENCH_DEFAULT_MAP_TASKS="80" */ Number of map tasks to generate data, read extended readme on selecting one * / export BIG_BENCH_DEFAULT_SCALE_FACTOR="10" / Scale Factor you would like to test 1000=1TB, 3000=3TB, 10000=10TB, 30000=30TB, 100000=100TB, 300000=300TB, 1000000=1PB/ export BIG_BENCH_DEFAULT_NUMBER_OF_PARALLEL_STREAMS="2" */Number of concurrent streams to use during throughput phase, default is 2 */ export BIG_BENCH_DEFAULT_BENCHMARK_PHASE="run_query" export BIG_BENCH_HADOOP_CONF="/etc/hadoop/conf.cloudera.hdfs" */ Adjust this to whatever distrubution of Hadoop you are using */ export BIG_BENCH_HADOOP_LIBS_NATIVE="/opt/cloudera/parcels/CDH/lib/hadoop/lib/native" */ Adjust this to whatever distrubution of Hadoop you are using */ export BIG_BENCH_DATAGEN_DFS_REPLICATION="1" / Use this setting to select number of replicas for your generated data on your HDFS file system, "1" means no replica, "3" is HDFS default of 3 copies. WARNING: This setting has no bearing on default HDFS replication for all other files, which is set to 3" export BIG_BENCH_STOP_AFTER_FAILURE="1" / the default behaviour is to stop when a query error occurs, set this to 0 to keep on running when an error occurs/
- Run $/bin/TPCxBB_Validation.sh */ Engine Validation Phase on SF1"
- Run $/TPCxBB_Benchmarkrun.sh */To execute the Benchmark, run twice to obtain "Performance" and "Repeatibility" numbers.
- Copy 3 set of logs from $../logs 1. From Engine Validation Phase logs. 2. Performance run logs. 3. Repeatability run logs. E.g logfile from one run:logs-20151029-135147-hive-sf1xxx.zip
- Engage the auditor and submit the report for publication.
Please contact bhaskar.gowda@intel.com for any questions.