Parseable is a fast observability platform on S3, built for high-performance log ingestion, querying, and analytics. By benchmarking Parseable against ClickBench, we showcase its ability to efficiently process structured and semi-structured data.

This benchmark utilizes Parquet files with an external table configuration to execute a series of analytical queries.

• AWS EC2 instance (recommended: c6a.metal or c6a.4xlarge)
• Canonical, Ubuntu, 24.04, amd64 noble image
• 500GB gp2 SSD root volume
• SSH access to the EC2 instance

1. Launch an AWS EC2 instance with the specification provided above.

2. Connect to the EC2 instance:

   ssh ubuntu@<instance-ip>
   

3. Clone the ClickBench repository:

   git clone https://github.com/ClickHouse/ClickBench
   cd ClickBench/parseable
   

4. Start the Parseable benchmark environment:

   bash benchmark.sh
   

Once Parseable is started, ingestion.sh is triggered which starts the ingestion process that performs the following tasks:

• Downloads and extracts hits.json.gz
• Splits the large JSON file into smaller files (2,500 lines each)
• Converts each file to proper JSON array format
• Creates a static schema in Parseable
• Ingests data in Parseable

Once data ingestion is complete, the benchmark queries are execute using run_query.sh This script will run a series of analytical queries against the ingested dataset and report performance metrics.