stardog-admin perf add2 Manual Page

NAME


stardog-admin perf add2 —Runs the add benchmark

SYNOPSIS


stardog-admin perf add2 {-d | --database} <Database> [ --ignore-errors ] [ --mode <Mode> ] [ {-p | --parallelism} <Parallelism> ] [ --pool <Pool> ] [ --random <Random> ] [ --raw-latency-recording ] [ --remove ] [ {-s | --stats} ] [ --server-side-files ] [ {-t | --transactions} <Transactions> ] --throughput <Throughput> [ {-tu | --output-time-unit} <Output Time Unit> ] [ {-w | --warmup} <Warmup> ] [--] [ <Files>... ]

OPTIONS


-d <Database>, --database <Database>
Stardog connection string. Embedded connection
--ignore-errors
Ignore errors encountered during transactions. Default: false.
--mode <Mode>
Test mode: append - Add to existing database, clear - First clear the contents of existing database, create - Create a new database, dropping it first if it already exists. Default: append.
-p <Parallelism>, --parallelism <Parallelism>
How many transactions are allowed to be in-flight at the same time. Default: 1
--pool <Pool>
Configure connection pooling. If set to 0 each transaction will open a new connection and close when it is done. If set to 1 a single connection will be reused for each thread. If set to greater than 1 a connection pool with that size will be used. Default value is 1, i. e. connections are reused per each thread.
--random <Random>
If any value (must be positive) is specified, this number of transactions or randomly generated triples is made in total. If this option is specified, input files must not be provided.
--raw-latency-recording
By default, so called 'corrected' latencies are recorded, assuming the expected latency of (1 / throughput) * parallelism, see the corresponding throughput and parallelism options. This option disables 'corrected' latency recording and falls back to 'raw' recording. See https://github.com/HdrHistogram/HdrHistogram#corrected-vs-raw-value-recording-calls for elaborate explanations of corrected vs. raw latency recording. Default: false.
--remove
Remove triples instead of adding them. Default: false.
-s, --stats
Enable auto statistics update (has effect only in create mode)
--server-side-files
When full files are added in transactions, assume that files reside on the database server side rather than the client side (where this `perf add` command is executed). Default: false.
-t <Transactions>, --transactions <Transactions>
The transaction size which is an integer that might be optionally followed by a letter 'f', 'k', or 'm'. The suffix 'f' is used to specify the number of files to use in each transaction (regardless of the file size). Otherwise, this option specifies the number of triples that will be added in each transaction where 'k' specifies thousands and 'm' specifies millions. The triples are taken from the provided files, or generated randomly, if --random option is specified.
--throughput <Throughput>
The target number of transactions per second that is going to be attempted to complete. If the actual throughput falls below this target, transactions are submitted at the maximum possible throughput, not exceeding the maximum specified parallelism. The actual observed throughput is printed afterwards. A required option.
-tu <Output Time Unit>, --output-time-unit <Output Time Unit>
The output time unit. Possible options: ns, us, ms, or s. Default: ms.
-w <Warmup>, --warmup <Warmup>
Transaction latency and throughput is not taken into account for this number of seconds since the beginning of the test. Default: 60 seconds.
--
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
<Files>
Files that will be added to the database or in random mode an integer that represents the number of triples in a transaction

DISCUSSION


Adds a set of files to an existing or newly created database and reports stastitics about each transactions .

EXAMPLES


Add each file in one transaction to an embedded server

    $ stardog-admin perf add -d http://local/lubm lubm/*.owl

Group 2 files into each transaction and add to a locally running server through http, clear the db first.

    $ stardog-admin perf add -t 2f -d http://localhost:5820/bsbm --mode clear bsbm10m/*.ttl.gz

Add 1 million triples in each transaction, recreate the database every time

    $ stardog-admin perf add -t 1m -d http://localhost:5820/perf --mode create bsbm100m/*.ttl.gz

Add 50K random triples in each transaction, do 100 total transactions

    $ stardog-admin perf add -t 50k -d http://localhost:5820/perf --random 100