The Dolt Timeline

REFERENCE
5 min read

One of the major themes here at DoltHub for the next year or so is trust. We built Dolt, the world's first version controlled SQL database. We know innovating at the database layer of your stack requires complete trust in the new technology you are adopting.

Trust is built over time. Many people don't realize the scope and length of Dolt's journey to a production quality Online Transaction Processing (OLTP) database. This timeline of major milestones in Dolt's history shows Dolt's journey to trustworthiness.

Dolt Timeline

2013-2018: Before Dolt

As hard as it is to believe, there was a time before Dolt existed. I'm not privy to the Noms folks musings on version controlled data but I had my first musings on the topic in July 2013.

Before Dolt Timeline

July 2013: Data as a Service

I was at an Amazon Web Services offsite. We were talking about potential new AWS services. I thought, first you could rent a computer. Then you could rent a database. Why can't you rent the data in the database? I called this idea data as a service and it kicked around in my head for the rest of my Amazon career through the duration of my next job.

June 2015: First Noms Commit

Noms is the technical predecessor to Dolt. Noms pioneered a new data structure called a Prolly Tree that makes version control of data at scale possible. In June 2015, the first Noms commit was made. A momentous day in the history of version controlled databases.

$ pwd
/Users/timsehn/dolthub/git/dolt
$ git log --reverse | head -6
commit 68c3ac02058e559367534aeeb7d9f8f483a4db1b
Author: Aaron Boodman <aaron@aaronboodman.com>
Date:   Tue Jun 2 20:45:33 2015 -0700

    first commit

August 2016: Noms Public Launch

A little over a year later, Noms was launched publicly.

January 2018: Attic Labs Acquired

A year and a half later, Attic Labs, the company building Noms, was acquired by Salesforce ceasing active development on Noms.

2018-2019: The Founding

I left my previous job with the intent of building a data sharing startup. I left on a Friday, started on the startup Monday, and by Thursday I decided I wasn't ready yet.

The Founding Timeline

February 2018: Liquidata Incorporated

But I kept kicking it along in the background and I got a new company called Liquidata incorporated in February 2018.

August 2018: Official Liquidata Founders Start Date

Aaron and Brian left their jobs in June and July and we were ready to start full time on Liquidata on August 6, 2018. Aaron and Brian came to my basement and we started kicking around ideas on how to build a data sharing startup, with the emphasis on version control. We were aware of Noms and thought the architecture was sound.

October 2018: Git Command Line Interface added to Noms

Most of our early focus was on what a website built for data sharing would look like. We also experimented with compiling and managing some datasets. Then, Brian put the Git command line on Noms, giving us "Git for Tables". We looked at that and decided, "This is a product. We should just ship this."

November 2018: New Tool Dubbed Dolt

In an homage to Git, we named this new tool Dolt. I won't go into why you would name a tool after a stupid person.

December 2018: Work on DoltHub Begins

Dolt needed a GitHub: a place on the internet to clone, push, and pull from. DoltHub was our answer to our data sharing mission.

2019: The Launch

We hired a few more people and went in the kitchen for almost a year. We emerged with Dolt, Git for Data, and DoltHub, GitHub for Dolt.

The Launch Timeline

August 2019: Dolt Launch

Dolt was launched on Aug 6, 2019 exactly a year after the founders got together in my basement.

September 2019: DoltHub Launch

Following quickly was the launch of DoltHub, a website to share Dolt databases, modeled after GitHub.

September 2020: Liquidata becomes DoltHub

After some initial success, we rebranded the company from Liquidata to DoltHub.

2020-2023: Data Sharing

The next chapter in DoltHub's history was a focus on building public databases. We first started to do this ourselves hoping to spur others into creating open databases.

Data Sharing Timeline

December 2020: Bounties Launch

We were building a nice supply of public, open databases but they weren't especially novel. We needed a way to show off the decentralized collaboration features of Dolt and DoltHub. We built a system to pay people for collecting and contributing data to Dolt databases called Data Bounties and launched it in December 2020 with an elections data bounty.

January 2022: DoltLab Launch

Though our focus was on open data sharing, potential customers kept asking for an on-premises DoltHub deployment. We answered their requests with the launch of DoltLab in January 2022.

September 2023: Bounties Ended

Dolt was starting to gain adoption as an OLTP database. To focus on becoming a database company, not a data sharing company, we ended the data bounties program.

2021-Present: OLTP

As hard as we tried to bootstrap a data sharing ecosystem, potential customers kept showing up asking if they could run Dolt as their production Online Transaction Processing (OLTP) database for various use cases. It felt like we were pushing data sharing but OLTP was pulling us.

OLTP Timeline

January 2021: First Production OLTP Customer

Our first production OLTP customer, Turbine, started testing Dolt as their production database in January 2021. They set some performance and reliability benchmarks for us to hit. We hit them and they are still a customer today.

June 2021: Announce OLTP Intentions

We announced our intentions to be an OLTP database in June 2021 after a bunch of OLTP work had been completed. At this point, Dolt already had secondary indexes, foreign keys, and all MySQL types. Transactions were behind a feature flag.

May 2022: Hosted Dolt Launch

Most OLTP databases make money from hosting. We think hosting Dolt for you is a potential money maker for us as well. We launched Hosted Dolt in May 2022.

August 2022: New Format Launched

In June 2021, the main unaddressed issue was performance. Dolt's storage format, inherited directly from Noms, was a bottleneck. We set out to build a new storage format that fit the performance and durability requirements of OLTP. We knew this would require old format Dolt databases to be migrated. We launched the first version of the new format in August 2022.

March 2023: New Format Complete

Six months later, the new format was complete. This set the stage for the 1.0 launch of Dolt.

2023-Present: 1.0

With the release of the new format combined with a number of other performance, correctness, and feature improvements, Dolt was production ready.

1.0 Timeline

May 2023: Dolt 1.0

In May 2023, we announced that Dolt was 1.0 and ready for your production use cases. Dolt had been running in production for two years at Turbine without issue.

November 2023: Doltgres Launch

Those familiar with the OLTP database space know, Postgres is the most popular SQL format. Dolt started MySQL-compatible. To capture part of the Postgres market, we started work on Doltgres, the Postgres-compatible version of Dolt in September 2023 and launched an Alpha version of Doltgres in November 2023.

The Future

What does the future hold for Dolt? We will continue to improve Dolt, fixing bugs in 24 hours or less. Doltgres will get production-ready, just like Dolt before it. We will continue to make improvements to Hosted Dolt, the primary way we intend to sell Dolt to the world. Interested in learning more? Come by our Discord and let's chat.

SHARE

JOIN THE DATA EVOLUTION

Get started with Dolt

Or join our mailing list to get product updates.