CockroachDB Videos
← Back to all videos

Why you should make your application multi-region from the start | Architectural Simplification

2024-04-04

Architectural Simplification

Description

In this video, Cockroach Labs Technical Evangelist Rob Reid takes a look at multi-instance architectures. When your application grows too quickly, it can be the side effect of a badly forecast organic but siloed growth of an application. To show what this looks like in action, Rob demos an application that has been deployed across the US, the EU, and Japan with a before and after scenario. 🔴 Before Scenario: Three siloed application architectures. One for the US. One for the EU. One for Japan. A US user will use the US connection string to talk to the database/application, etc. This setup may result in configuration creep, which is the gradual unintended divergence of configuration between applications along with divergence of code and database. This can become difficult to manage over time. 🟢 After Scenario: One instance of the architecture, with geo-aware load-balancing and geo-partitioning. One instance of the system of record and application to update with zero risk of config drift. A few benefits of the After scenario ✔️ Low architectural complexity ✔️ Low operational complexity ✔️ Zero config-creep between regions ✔️ Better resource utilization ✔️ More resilient to failure GitHub: https://cockroa.ch/3Q090Fd Further reading ✔️ What is CockroachDB? https://cockroa.ch/42uaZ9G ✔️ CockroachDB architecture overview: https://cockroa.ch/3RzIvrh ✔️ Global tables in CockroachDB: https://cockroa.ch/440S2MV Watch more videos in our Architectural Simplification series: https://www.youtube.com/playlist?list=PL_QaflmEF2e9Dgiw8lW-Z8jq7TNDNy_3V Why you should make your application multi-region from the start | Architectural Simplification 00:00 Introduction 00:14 Comparing the Before and After scenarios 01:06 Testing the Before scenario 03:48 Real-world problems with siloed application architectures 05:14 Testing the After scenario 08:28 What does a global tables topology pattern look like in terms of real-world performance? 10:05 Pinning market-specific product data and joining onto global tables 10:56 Why you should consider multi-region architecture