stardog-admin perf query Manual Page

NAME


stardog-admin perf query —Run the query performance benchmark

SYNOPSIS


stardog-admin perf query [ --counts ] {-d | --database} <Database> [ --exclude-outliers ] [ {-g | --graph} <Query> ] [ {-m | --memory} ] [ --memory-management ] [ --mode <Mode> ] [ --multi-query ] [ {-p | --passwd} <password> ] [ {-r | --reasoning} ] [ {-s | --skip} <Queries to skip>... ] [ {-t | --trials} <Trials> ] [ {-u | --username} <username> ] [ {-v | --verbose} ] [ {-w | --warmups} <Warmups> ] [ --waterloo ] [--] <Queries>...

OPTIONS


--counts
Run queries only once and print the number of results each query returns. If this flag is set all performance related options are ignored
-d <Database>, --database <Database>
Stardog database to query. This is either a connection string or the path to an index directory
--exclude-outliers
Exclude the min and max times for each query when computing statistics (Requires at least 3 trials).
-g <Query>, --graph <Query>
Default graph to run the queries against
-m, --memory
Profile memory usage of queries. Memory profiling is approximate and based on the heap/non-heap size reported by the JVM. Profilinf relies on the assumption that we can trigger a manual GC by repeatedly asking GC and checking if memory usage is decreased.
--memory-management
Enable memory management over direct index connections
--mode <Mode>
Connection mode (Direct, Sesame, SesameStardog, Jena, JenaStardog)
--multi-query
Flag to indicate the input file has multiple SPARQL queries, one query per line.
-p <password>, --passwd <password>
Password.
-r, --reasoning
Enable reasoning
-s <Queries to skip>, --skip <Queries to skip>
Skips the queries with given name.
-t <Trials>, --trials <Trials>
The number of test trials to run, the default is twenty (20).
-u <username>, --username <username>
User name.
-v, --verbose
Verbose output
-w <Warmups>, --warmups <Warmups>
The number of warmups to run, the default is five (5).
--waterloo
Waterloo stress testing
--
This option can be used to separate command-line options from the list of argument, (useful when arguments might be mistaken for command-line options
<Queries>
Queries to run.

DISCUSSION


Executes a benchmark; for each query, a specified number of warm-up runs are executed and then a set number of trial executions are run. The time per execution for those trials is recorded and at the end of the benchmark, a short report is generated with the results of the benchmark. The report is saved into the current directory.

EXAMPLES


Execute the benchmark against an embedded server

    $ stardog-admin perf run -d embed://lubm1m test/data/eval/lubm/queries/*.sparql

Execute the benchmark with a given number of warm-ups and trial runs plus reasoning on a locally running server.

    $ stardog-admin perf run --warmups 25 --trials 100 -d http://localhost:8080/lubm1m --reasoning test/data/eval/lubm/queries/query?.sparql

Execute the benchmark against an index directly, skipping some queries

    $ stardog-admin perf sp2b --skip q4 q5a -d file:/stardog_home_path/sp2b5m profiler/queries/sp2m/*.sparql