July 2020 Summaries

In May, Supabase launched and quickly gained popularity on platforms like Hacker News and GitHub. During the launch week, they had over 30,000 new website visitors, more than 1400 signups, and a significant increase in their GitHub stars. The middleware served from a single Ubuntu server with docker-compose up survived well, while the frontend was hosted on Netlify and Vercel, which were both reliable. However, they faced issues with Digital Ocean's cloud limits and Cloudflare's subdomain limit during the launch. They also experienced production errors from Digital Ocean and had to migrate 1800 servers over to AWS due to their frustrating credits policy. Supabase is now on the AWS "Activate" program, which grants them $100,000 in credits, allowing for improvements in database startup times and a better experience for developers.

Here are the key points about Supabase's launch and growth: Supabase recently hit the top of Hacker News, stayed on the front page for over 24 hours, and has since been featured on the Stack Overflow podcast and scaled to over 1000 databases. The company's website saw a surge in visitors with 30,000 new site visitors and over 1400 signups in just seven days. Despite some initial challenges, Supabase's middleware and frontend services were surprisingly resilient, with the middleware being served from a single server that was initially intended for alpha testing. However, the company faced issues with cloud providers Digital Ocean, including hitting server limits and experiencing production errors, which ultimately led to a migration to AWS, where they are now utilizing the t3a instances, which have improved database startup times significantly.

PostgreSQL Templates are standard system databases that exist in every newly created database cluster, including template1 and template0. By default, new databases are created from the template1 database, which can be modified to include custom tables, data, extensions, or procedural languages. However, it is not advisable to modify template1 directly as any changes made would need to be manually uninstalled or dropped. Instead, users can create their own custom template databases and set them as templates for new databases. The template0 database serves as a fallback in case anything irreversible happens to template1. It contains the same data as template1 but should never be modified after the database cluster has been initialized.

When creating a new database in Postgres, you're essentially basing it on an existing standard system database called `template1`, which contains all the default settings and configurations for a new database cluster. This means that any modifications made to `template1` will be automatically applied to subsequent databases created from it. However, this also means that if something goes wrong with `template1`, you'll have to recreate it from scratch. To avoid this, you can create a custom template database using the `ALTER DATABASE` command and base new databases from it instead. This allows for more control over the settings and configurations of your database cluster, and provides a safe way to create databases without worrying about polluting the default `template1`. Additionally, if something goes wrong with `template1`, you can fall back on the `template0` database, which contains the same data as `template1` but should never be modified. By using custom template databases, you can create a clean and organized database cluster that meets your specific needs.

PostgreSQL backups can be categorized into two types: logical and physical. Logical backups involve translating all the data into a set of SQL commands and writing it into a single file, which can then be fed to any database cluster to recreate everything. Physical backups pertain to the actual set of files or file systems where your database data is stored. Each type has unique uses depending on the needs of the user: Logical backups are simpler for smaller databases and better for handling larger ones, while physical backups are more ideal for larger databases and can be used for point-in-time recovery.

Logical and physical backups are two types of methods used to backup PostgreSQL databases, each with its own advantages and use cases. Logical backups involve translating data into SQL commands that can be executed on any database cluster, making them ideal for simple backups, migration between major versions, and backing up a single database. On the other hand, physical backups involve copying actual files or file systems where database data is stored, making them more suitable for larger databases and achieving point-in-time recovery through Write Ahead Log (WAL) files and tools like WAL-G.

In June 2020, Supabase released a major update with several new features and improvements. These include an excel-like interface for managing relational data, easy handling of JSON data in Postgres, the ability to choose your database region, and daily backup exposure on the dashboard. The company also welcomed Steve Chavez as a team member and announced plans to hire a part-time designer or UX expert. Supabase is focused on making Postgres as simple as Firebase while maintaining zero lock-in principles.

We're now 4 months into building Supabase, which means another major update. We've released an excel-like editing interface that can drill down into relational data, making Postgres even easier to use than document stores. We've also improved the management of JSON data and added support for choosing a database region, allowing users to go multi-region with instant replication close to their customers. Additionally, we've exposed daily database backups on the dashboard for simple migration purposes, and made various small improvements to the SQL editor, API docs, UX, and hiring process. Steve Chavez has joined our team as a maintainer of PostgREST, and we're currently hiring a part-time designer or UX expert.