Reinventing the Database Stack: The Story and Innovation of 8kdata

In the fast-paced world of data engineering, architects have long faced a persistent architectural dilemma: choose the rigid structure and rock-solid ACID guarantees of traditional relational databases like PostgreSQL, or embrace the flexible, schema-less agility of document-oriented NoSQL databases like MongoDB.

During the mid-2010s, a Spanish database research and development company named 8Kdata stepped into this divide with a bold vision. Rather than treating relational and non-relational systems as mutually exclusive, 8Kdata built innovative middleware designed to blend the best of both worlds.

The Genesis of 8Kdata

Founded by database veterans and open-source advocates (including notable PostgreSQL contributors like Álvaro Hernández Tortosa), 8Kdata positioned itself as an R&D powerhouse. Operating primarily out of Spain, the company focused entirely on solving complex database engineering challenges.

Instead of building a brand-new storage engine from scratch—a notoriously difficult and risky endeavor in the data space—8Kdata looked at leveraging existing giants. Specifically, they targeted PostgreSQL, widely lauded for its extensibility, resilience, and industrial-grade reliability, and asked a radical question: What if Postgres could speak MongoDB?

ToroDB: The Flagship Innovation

The crowning achievement of 8Kdata was ToroDB, an open-source project designed to bridge the gap between document-oriented design and relational execution.

ToroDB functioned as a middleware layer that allowed developers to interact with a database using MongoDB’s familiar drivers, query language, and the Mongo Wire Protocol, while utilizing PostgreSQL as the underlying storage and execution layer.

Key Features of the ToroDB Architecture:

• MongoDB Compatibility: Applications could connect to ToroDB using standard MongoDB drivers without altering client-side code.
• Relational Backbone: Unlike native NoSQL engines that often sacrificed consistency for availability, ToroDB mapped documents into relational tables under the hood, allowing users to leverage PostgreSQL’s robust ACID transactions and data durability.
• Custom Protocols (mongowp): To achieve this, 8Kdata engineered foundational libraries like mongowp, a Go/Rust/Java-adjacent wire protocol parser that helped other developers easily interpret MongoDB traffic.

Through this approach, 8Kdata tackled a major pain point for engineering teams: getting the rapid prototyping speed of JSON document stores without abandoning the mature management, backup, and querying toolsets built around relational databases.

Expanding the Ecosystem: ToroDB Stampede

Recognizing that organizations often struggled with data migration and cross-platform analytics, 8Kdata also introduced secondary utilities like ToroDB Stampede.

Stampede was built to solve a very specific operational challenge: replicating data in real time from a MongoDB replica set directly into a structured SQL database. This allowed enterprises to feed NoSQL operational data straight into traditional Business Intelligence (BI) tools (such as Tableau or Pentaho) that natively preferred structured relational schemas. In effect, it served as a seamless bridge for companies transitioning between architectural paradigms or seeking unified reporting.

The Legacy and Impact

Though 8Kdata’s core open-source efforts eventually wound down toward the end of the 2010s—coinciding with PostgreSQL introducing its own native jsonb data types and built-in JSON indexing capabilities—the company’s footprint on the database community remains significant.

8Kdata anticipated a trend that defines modern data architecture today: multimodel databases. Modern data platforms rarely fit neatly into a single silo; instead, they blend relational consistency with document, vector, and time-series flexibility.

By pushing the boundaries of what could be achieved by coupling middleware with PostgreSQL, 8Kdata proved that imagination and deep systems-level engineering could challenge dogmatic divides in software development.