Treasure Data

Treasure Data | NotebookLM

Metrics

In 2023, Treasure Data currently serves over 700 accounts and more than 6,000 users. Since 2022, platform usage has doubled. Each day, the system handles 2.5 million+ Trino queries and over 100,000 Hive queries. It processes more than 200 trillion rows daily and performs over 10 billion S3 GET requests per day to read partition data from AWS S3.

Overview

How Treasure Data reduced ad spend and increased marketing ROI by 20%

Treasure Data Enterprise Customer Data Platform | Demo

Tech Stack

• Wvlet, similar to BigQuery's pipe query syntax
• Trino
• Hive 4 (Hive 4.0.x comes with Iceberg 1.4.3 included.)
• Plazma (Closed Source)
  • PostgreSQL (metadata)
  • S3
    • Real-time Storage
    • Archive Storage
• fluentd
• embulk: Data Connector and Result Export

LinkedIn Post

Data Architecture

Data Architecture 2021 (Source: Secure exchange SQL - Treasure Data at Trino Summit 2023)

Data Architecture 2023 (Source: Secure exchange SQL - Treasure Data at Trino Summit 2023)

Pros and Cons (Source: What is Treasure Data CDP?)

Zero-Copy Integrations Between Your CDP and Data Warehouse

Data Ingestion

• Google Analytics Data API Import Integration

Questions

• Data Lakehouse Architecture? proprietary (MPC1)
• How to deal with dbt Core? (dbt Fusion)
• SQLMesh

Amazon S3 Parquet Export Integration Amazon Elastic MapReduce Integration

https://www.linkedin.com/in/pramodmanjappa/