Skip to content

Latest commit

 

History

History
64 lines (53 loc) · 2.68 KB

README.md

File metadata and controls

64 lines (53 loc) · 2.68 KB
Raw

Node.js CrateDB Ingest Benchmark

A multi-process Node.js script to run high-performance ingest benchmarks on CrateDB clusters. The script generates random data and runs batched insert statements against a single table using CrateDB's HTTP endpoint.

The top measured performance (single Node.js process, seven-node CrateDB cluster with 30 vCPUs each) was 1,200,000 rows/s.

Setup

  1. Install Node.js using one of the available installation methods
  2. Clone this repository: git clone https://github.com/proddata/nodeIngestBench.git
  3. Change into the cloned repository: cd nodeIngestBench
  4. Install dependencies: npm install
  5. Configure the connection to your CrateDB cluster by creating a .env file:
CRATE_HOST=localhost
CRATE_USER=admin
CRATE_PASSWORD=secret-password
CRATE_PORT=4200
CRATE_SSL=true

Running benchmarks

Start the benchmarking with node appCluster.js. The script takes several optional command-line parameters:

  • table: The name of the table used for benchmarking. The table will automatically be created if it doesn't exist yet.
  • shards: The number of shards that will be allocated for the table.
  • replicas: The number or range of replicas to use for the table.
  • extra_tags_length: The number of extra values added to the tags object in the table. This can be used to generate a wider table with more columns.
  • drop_table: If true, the table will be dropped and re-created when running the script.
  • batch_size: The number of rows that are inserted as part of a single INSERT statement.
  • processes: The number of child processes that will be spawned. Each child process inserts data independently.
  • concurrent_requests: Each child process will run this number of concurrent queries.
  • max_rows: The number of rows after which a child process will terminate. Overall, the (maximum) number of rows that will be inserted is processes * max_rows.

Example

Below is a basic example that was run on a three-node CrateDB cluster with 10 vCPUs each:

$ node appCluster.js --batch_size 20000 --max_rows 1000000 --shards 12 --concurrent_requests 20 --processes 1

-------- Options ---------
{
  dropTable: true,
  processes: 1,
  batchSize: 20000,
  maxRows: 6000000,
  table: 'doc.cpu',
  shards: 6,
  concurrentRequests: 20,
  extraTagsLength: 0,
  replicas: 0
}
--------------------------
[...]
-------- Global Results ---------
Time	 25.179 s
Rows	 6,020,000 records
Speed	239,088.13 rows per sec
---------------------------------

As each row contains ten numeric metrics, this is equal to a throughput of 2,400,000 metrics/s.