The HLR benchmark application is a Java application that interacts with the database using JDBC. The application is highly tuned to deliver optimal performance in accordance with database programming best practices. It also implements comprehensive performance measurements and reports a wide variety of metrics such as throughput (TPS), minimum, maximum, average, and 90th percentile transaction response times.

The online help for the application is as follows:

Usage:

java HLR [options]

     -c[lient]            connect in client mode (TimesTen only)

     -mysql               use MySQL rather than TimesTen

     -ora[cle]            use Oracle rather than TimesTen

     -sub[scribers] n     number of prepopulated subscribers

     -minsub[scriber] n   minimum subscriber id to target

     -maxsub[scriber] n   maximum subscriber id to target

     -h[ost] n            host for MySQL or Oracle}

     -p[ort] n            port number for MySQL or Oracle}

     -s[id] n             SID for Oracle

     -uid n               user name for TimesTen or Oracle

     -pwd n               password for TimesTen or Oracle

     -r[ate] n            number of ops/sec to run

     -t[hreads] n         number of threads to use

     -ti[me] n            time limit; run for n seconds

     -d[etail]            print per-thread measurements

     -tr[ace]             turn on jdbc tracing

     dsn=<mydsn>          specifies DSN for target database (TimesTen only)

Examples:

  (TimesTen) java HLR -o 1000000 -ti 300 -t 8 -uid scott -pwd tiger dsn=sampledb

  (MySQL)    java HLR -mysql -host host1 -port 12345 -t 8

  (Oracle)   java HLR -ora -host host1 -p 1521 -s orcl -uid scott -pwd tiger -t 8

The actual command lines and parameters used for each scenario in this study are shown below. We varied the number of threads for the two use cases, as described in Use cases, test methodologies, and results.

Here is an example of the HLR command that we ran during the baseline 400M subscribers RAC-only tests:

java HLR -oracle -host tt-orac-n1 -port 1521
         -sid ttoracdb.domain.com -uid appuser -pwd appuser
         -sub 2500000000 -minsub 1 -maxsub 200000000
         -threads 32 -rup 120 -rdown 60 -time 300
         -csv ora400_1.csv

java HLR -oracle -host tt-orac-n2 -port 1521
         -sid ttoracdb.domain.com -uid appuser -pwd appuser
         -sub 2500000000 -minsub 1250000001 -maxsub 1450000000
         -threads 32 -rup 120 -rdown 60 -time 300
         -csv ora400_2.csv

Here is an example of the HLR command that we ran during the baseline 1,800M Subs RAC-only tests:

java HLR -oracle -host tt-orac-n1 -port 1521
         -sid ttoracdb.domain.com -uid appuser -pwd appuser
         -sub 2500000000 -minsub 1 -maxsub 900000000
         -threads 1 -rup 120 -rdown 60 -time 300
         -csv ora1800_1.csv

java HLR -oracle -host tt-orac-n2 -port 1521
         -sid ttoracdb.domain.com -uid appuser -pwd appuser
         -sub 2500000000 -minsub 1250000001 -maxsub 2150000000
         -threads 1 -rup 120 -rdown 60 -time 300
         -csv ora1800_2.csv

Here is an example of HLR command that we ran during the 400M subscribers TimesTen tests:

java HLR -rate 220000 -sub 400000000 -uid appuser -pwd appuser
         -threads 4 -rup 120 -rdown 60 -time 300 -csva tt400.csv
         dsn=sampledb

Here is an example of the HLR command that we ran during the 1,800M subscribers TimesTen tests: