Cockroach Labs Blog

CockroachDB 1.0

Launching CockroachDB 1.0, a production-ready release with distributed SQL and multi-active availability in two tiers: CockroachDB Core and CockroachDB Self-hosted.

Relicensing CockroachDB

Why are my Go executable files so large?

Golang executable files get too large for reasons that are not relevant to server software with long-running processes, like CockroachDB, and its incurred space cost is particularly inconvenient for large, feature-rich programs.

CockroachDB license change

What you need to know about the changes that happened to CockroachDB Enterprise this past November.

CockroachDB 2.0 Performance Makes Significant Strides

Learn which performance metrics matter, why to demand benchmark TPC-C results under serializable isolation, and how CockroachDB 2.0’s performance compares to 1.1.

Hello World

We started Cockroach Labs with a clear objective: build the database we'd been itching after for years, and build it as open source, from the start. Today, we’re launching CockroachDB for everyone.

CockroachDB 2.0 released

Since our 1.0 release, we have been hard at work making CockroachDB 2.0 the database that enables a company’s data layer to evolve with their business -- whether they are in a single datacenter or expanding around the world.

2022 Cloud Report

The 2022 Cloud Report is the only cloud performance report to compare AWS, Azure, and GCP on benchmarks that reflect critical applications and workloads.

Pebble: A RocksDB Inspired Key-Value Store Written in Go

Today we’re introducing Pebble, a RocksDB-inspired and RocksDB compatible key-value store written in Go that brings greater stability to CockroachDB.

GCP Outpaces Azure, AWS in the 2021 Cloud Report

Now in its third year, the 2021 Cloud Report from Cockroach Labs compares AWS, Azure, and Google (GCP) to tell realistic performance stories on behalf of mission-critical OLTP applications.

2020 Cloud Report: AWS vs. GCP vs. Azure

The 2020 Cloud Report compares AWS, GCP, and Azure on microbenchmarks and customer-like-workloads to understand performance tradeoffs within each cloud.

How we built a serverless SQL database

CockroachDB serverless is the first truly scalable serverless SQL database. Learn how it works and how a serverless sql database can relieve the operational burden of developers.

My Go executable files are still getting larger

The original draft of the analysis we presented was reviewed internally, and yet still inaccuracies slipped through. We are a team of humans. We make mistakes. We take full responsibility for the invalid conclusions we published.

AWS, Azure, and GCP respond to cloud report

Key learnings about how to get the best performance from GCP, AWS, and Azure based on original research and conversations with the big three.

A Brief History of High Availability

The perennial question of homo sapiens is, 'How did we get here?' Today we're going to take a crack at answering that: where 'here' is defined as 'high availability for web services'.

Five months treating bugs like patients and coding agents like a medical team

How Cockroach Labs built MOLT Sinai, an AI coding “teaching hospital” that shipped IBM Db2 migration support in under two days.

AWS Outperforms GCP in the 2018 Cloud Report

Customers rely on us to help navigate the increasingly competitive cloud wars. This inspired the 2018 Cloud Report, a benchmark of AWS and GCP on performance, latency, CPU, network, I/O, and cost.

Raft Is So Fetch: The Raft Consensus Algorithm Explained Through Mean Girls

Understanding the Raft Consensus Algorithm can be tough. In honor of "Mean Girls" day (October 3), we present Raft as explained through the movie "Mean Girls."

Distributed Databases Should Work More Like CDNs

Distributed database performance––both latency and uptime––can be dramatically improved with CDN-like multi-region deployments.

How We Built a Vectorized SQL Engine

Using vectorized processing in an execution engine makes more efficient use of modern CPUs by changing the data orientation (from rows to columns) to get more out of the CPU cache and deep instruction pipelines by operating on batches of data at a time.

Sqlfmt: an opinionated online SQL formatter

sqlfmt is an opinionated SQL formatter. It’s not designed to be completely customizable, and we hope to remove options over time. Use it and stop thinking about how to format SQL.

CockroachDB Stability Post-Mortem: From 1 Node to 100 Nodes

In August, we wrote about our concerted stability effort. Today, we've more than achieved our CockroachDB stability goal of running a 100-node cluster.

How We Built a Cost-Based SQL Optimizer

CockroachDB 2.1 includes a brand-new, built-from-scratch, cost-based SQL optimizer. This post explains what a cost-based SQL optimizer is, and tells the story of how we decided we really, really needed one.

Parallel Commits: A New Atomic Commit Protocol for Distributed Transactions

Parallel Commits are an atomic commit protocol in CockroachDB. The feature promises to halve the latency of distributed transactions by performing all consensus round trips required to commit a transaction concurrently.

The Limits of the CAP Theorem

This post explains how CAP-Consistent systems can be highly available, how the CAP theorem applies to CockroachDB, and why consistency is the better choice.

Building CockroachDB on top of RocksDB

CockroachDB uses RocksDB for its storage engine because of RocksDB's rich feature set, which is necessary for a complex product like a distributed SQL database.

CockroachDB 1.1 Released: Production Made Easy

CockroachDB 1.1 includes support for faster migrations from legacy databases, simplified cluster management, and improved performance.

Why CockroachDB and PostgreSQL Are Compatible

CockroachDB is PostgreSQL compatible in order to be accessible to an ecosystem of drivers, existing code, and developer knowledge.

How We’re Building a Business to Last

Avoid Vendor Lock-In Risk with Multi-Cloud Deployments

Using CockroachDB means shifting operational thinking from disaster recovery to disaster resilience, and completely side steps the risks associated with vendor lock-in.