JBR Experiment - LDBC SNB Decentralized

A jbr experiment type for the decentralized LDBC SNB social network benchmark.

Requirements

• Node.js (1.12 or higher)
• Docker (required for invoking LDBC SNB generator)
• jbr (required for initializing, preparing, and running experiments on the command line)

Quick start

1. Install jbr

jbr is a command line tool that enables experiments to be initialized, prepared, and started. It can be installed from the npm registry:

$ npm install -g jbr

or

$ yarn global add jbr

2. Initialize a new experiment

Using the jbr CLI tool, initialize a new experiment:

$ jbr init ldbc-snb-decentralized my-experiment
$ cd my-experiment

This will create a new my-experiment directory with default configs for this experiment type.

3. Configure the required hooks

This experiment type requires you to configure a certain SPARQL endpoint to send queries to for the hookSparqlEndpoint. A value for this hook can be set as follows, such as sparql-endpoint-comunica:

$ jbr set-hook hookSparqlEndpoint sparql-endpoint-comunica

4. Prepare the experiment

In order to run all preprocessing steps, such as creating all required datasets, invoke the prepare step:

$ jbr prepare

All prepared files will be contained in the generated/ directory.

5. Run the experiment

Once the experiment has been fully configured and prepared, you can run it:

$ jbr run

Once the run step completes, results will be present in the output/ directory.

Output

The following output is generated after an experiment has run.

output/query-times.csv:

name;id;results;time;timestamps
interactive-short-4;0;0;7;
interactive-short-4;1;0;5;
interactive-short-4;2;0;6;
interactive-short-4;3;0;3;
interactive-short-4;4;0;3;
interactive-short-5;0;0;0;
interactive-short-5;1;0;0;
interactive-short-5;2;0;0;
interactive-short-5;3;0;0;
interactive-short-5;4;0;0;

output/stats-server.csv: Per second of the experiment: CPU percentage, memory usage (bytes), memory percentage, received bytes, transmitted bytes.

cpu_percentage,memory,memory_percentage,received,transmitted
9.915362228116711,10489856,0.5024267940030527,488,0
9.863725050505051,17354752,0.8312308965993495,648,0
9.64850952141058,19116032,0.915589944401502,738,0
9.345685076142132,23072768,1.105103526208198,738,0
10.029959365079364,26759168,1.2816689750964243,738,0
10.25411566137566,30363648,1.45431074734269,738,0

output/logs/server.txt: Logs of the Solid Community Server.

Configuration

The default generated configuration file (jbr-experiment.json) for this experiment looks as follows:

{
  "@context": [
    "https://linkedsoftwaredependencies.org/bundles/npm/jbr/^1.0.0/components/context.jsonld",
    "https://linkedsoftwaredependencies.org/bundles/npm/@jbr-experiment/ldbc-snb-decentralized/^1.0.0/components/context.jsonld"
  ],
  "@id": "urn:jrb:my-experiment",
  "@type": "ExperimentLdbcSnbDecentralized",
  "scale": "0.1",
  "configGenerateAux": "input/config-enhancer.json",
  "configFragment": "input/config-fragmenter.json",
  "configFragmentAux": "input/config-fragmenter-auxiliary.json",
  "configQueries": "input/config-queries.json",
  "configServer": "input/config-server.json",
  "directoryQueryTemplates": "input/templates/queries",
  "overwriteFilesDuringPrepare": false,
  "hadoopMemory": "4G",
  "dockerfileServer": "input/dockerfiles/Dockerfile-server",
  "serverPort": 3000,
  "serverLogLevel": "info",
  "serverResourceConstraints": {
    "@type": "DockerResourceConstraints",
    "cpu_percentage": 10
  },
  "endpointUrl": "http://localhost:3001/sparql",
  "queryRunnerReplication": 3,
  "queryRunnerWarmupRounds": 1,
  "queryRunnerRecordTimestamps": true,
  "hookSparqlEndpoint": {
    "@id": "urn:jrb:cc:hookSparqlEndpoint",
    "@type": "HookNonConfigured"
  }
}

Any config changes require re-running the prepare step.

More background information on these config options can be found in the README of the decentralized LDBC SNB social network benchmark.

Configuration fields

• scale: The SNB scale factor
• configGenerateAux: Path to enhancement config for ldbc-snb-enhancer.
• configFragment: Path to fragmentation config for rdf-dataset-fragmenter.
• configFragmentAux: Path to enhancement's fragmentation config for rdf-dataset-fragmenter.
• configQueries: Path to query instantiation config for sparql-query-parameter-instantiator.
• configServer: Path to server config for Solid Community Server.
• directoryQueryTemplates: Path to the directory containing SPARQL query templates for sparql-query-parameter-instantiator.
• overwriteFilesDuringPrepare A flag to indicate whether or not the fragmentations should always be fully recreated during the prepare step. Otherwise they will be cached until deleted manually.
• hadoopMemory: Memory limit for Hadoop for LDBC SNB.
• dockerfileServer: Path to a Dockerfile for Solid Community Server.
• serverPort: HTTP Port for Solid Community Server.
• serverLogLevel: Logging level for Solid Community Server.
• serverResourceConstraints: Resource constraints for the Solid Community Server Docker container.
• endpointUrl: URL through which the SPARQL endpoint of the hookSparqlEndpoint hook will be exposed.
• queryRunnerReplication: Number of replication runs for sparql-benchmark-runner.
• queryRunnerWarmupRounds: Number of warmup runs for sparql-benchmark-runner.
• queryRunnerRecordTimestamps: Flag to indicate if result arrival timestamps must be recorded sparql-benchmark-runner.

License

jbr.js is written by Ruben Taelman.

This code is copyrighted by Ghent University – imec and released under the MIT license.