Geonames workload

This workload is based on a geonames dump of the file allCountries.zip retrieved as of April 27, 2017. This is a fixed historical snapshot — it is not updated as the live Geonames database changes — which ensures reproducible benchmark results across runs and time.

For further details about the semantics of individual fields, please see the geonames dump README.

Modifications from the original data:

• The original CSV data have been converted to JSON.
• The original longitude and latitude fields are combined into a single location field ([longitude, latitude] array in GeoJSON order).

Example Document

{
  "geonameid": 2986043,
  "name": "Pic de Font Blanca",
  "asciiname": "Pic de Font Blanca",
  "alternatenames": "Pic de Font Blanca,Pic du Port",
  "feature_class": "T",
  "feature_code": "PK",
  "country_code": "AD",
  "admin1_code": "00",
  "population": 0,
  "dem": "2860",
  "timezone": "Europe/Andorra",
  "location": [
    1.53335,
    42.64991
  ]
}

Test Procedures

• append-no-conflicts (default): Indexes the whole corpus with unique document IDs, then runs a suite of search and faceting queries.
• append-no-conflicts-index-only: Indexes the whole corpus with unique document IDs, no queries.

Parameters

This workload allows the following parameters via --workload-params:

Indexing

• bulk_size (default: 5000): Number of documents per bulk indexing request.
• bulk_indexing_clients (default: 8): Number of clients that issue bulk indexing requests.
• warmup_time_period (default: 120): Seconds of indexing warm-up before measurements begin.
• ingest_percentage (default: 100): A number between 0 and 100 that defines how much of the document corpus should be ingested.
• conflicts (default: "random"): Type of id conflicts to simulate. Valid values are sequential (a document id is replaced with a sequentially increasing id) and random (replaced with a random other id).
• conflict_probability (default: 25): A number between 0 and 100 that defines the probability of id conflicts. Only applies when running index-update operations.
• on_conflict (default: "index"): Whether to use an "index" or "update" action when simulating an id conflict.
• recency (default: 0): A number between 0 and 1 that biases towards more recent ids when simulating conflicts. 0 means uniform distribution; 1 means always pick the most recent id.
• error_level (default: "non-fatal"): Error handling for bulk operations. Use "fatal" to abort on any indexing error.

Collection

• num_shards (default: 1): Number of shards for the Solr collection.
• replication_factor (default: 1): Number of NRT replica copies per shard.

Optimize

• max_segments (default: 1): Target number of segments after force-merge (optimize).
• optimize_timeout (default: 600): Timeout in seconds for the optimize operation.

Search

• target_throughput: Global default for maximum requests per second across all search operations. Use none for no limit. Individual operations can be overridden with <operation_name>_target_throughput.
• search_clients (default: 1): Number of clients issuing search requests.
• warmup_iterations: Global default warmup iterations before measurement begins.
• iterations: Global default number of measured iterations per operation.

License

The data originates from GeoNames and is licensed under the Creative Commons Attribution 3.0 License. Use of this data requires crediting GeoNames as the source.

This work is licensed under a Creative Commons Attribution 3.0 License,
see http://creativecommons.org/licenses/by/3.0/
The Data is provided "as is" without warranty or any representation of accuracy, timeliness or completeness.