Community chose PlanetScale for flexibility, scale, and speed.
Community originally chose Cassandra as their database, taking advantage of NoSQL’s un-opinionated data structure to get their business off the ground. When Community began to plan for the scale of sending millions of messages a month they knew they were going to need to find a relational database that gave them the same scale and flexibility of a NoSQL database. By migrating to PlanetScale as their database in the cloud, Community was able to get that scale and flexibility while continuing to scale their business.
Planning for scale and finding the limitations of NoSQL
Community was founded in 2019 as a way to connect brands, celebrities, and enterprise companies directly to their audiences via text messaging. They subvert the typical “marketing” strategies taken by celebrities by using technology to directly connect brands and personalities with their millions of fans and customers. By utilizing text messaging they have seen a greater increase in audience engagement compared to typical communication channels like emails and social media. “We grew pretty rapidly through word of mouth and early success with notable celebrities,” says Karl Matthias, Director of Architecture and Platform Engineering at Community.com. The team had two problems: rapidly building the platform while simultaneously scaling the platform.
Originally the team at Community chose Apache Cassandra, but quickly realized they needed more flexibility in their data layer.
The engineering team at Community identified two limitations of their existing NoSQL database:
Lack of flexibility
They couldn’t rewrite queries easily without changing the way the data was written
Slow speed of development
Often, new features required rewriting some or all of the data
“The last thing we wanted to pay for was an engineering team to host our own [database], especially when we were trying to get a product in the market. You don’t need to hire engineers to manage databases if you can pay somebody else to do it.”
“We knew we were going to have to have a very scalable store for messages, because we were going to send billions of messages,” says Karl.
Once they began cataloguing their requirements, Community landed on a short but critical list for their database:
- Easily change schemas and join on new tables
- Able to shard indefinitely
- Hosted, no need to manage infrastructure
Results: Speed and control
“The truth is that sharding is an art. You can’t know all of the requirements you will eventually have up front, so you have to think it through carefully. Having the flexibility to change our sharding strategy later is a huge win for us.”
Luckily the team had heard of Vitess and PlanetScale and found that it would give them both sharding and a reliable key store that would allow them to change data structure on the fly.
Once Community had migrated to PlanetScale, they saw an immediate benefit in their engineering workflow and ability to ship quicker. “I used to work for New Relic and it was nice to work on a flexible SQL store that can scale—without building and operating all of the scaling tech ourselves, like we did back then.” says Karl after switching to PlanetScale’s cloud database.
One benefit the team appreciated from the start was not needing to stand-up their own database cluster. “The last thing we wanted to pay for was an engineering team to host our own [databases], especially when we were trying to get a product in the market. You don’t need to hire engineers to manage databases if you can pay somebody else to do it.” Aside from not worrying about infrastructure, the team is also looking forward to the upcoming release of developer features that will make schema changes and migrations even easier.
The second immediate benefit Community saw from switching to PlanetScale was having explicit control of their sharding story. Unlike many NoSQL cloud data providers, sharding with PlanetScale is not a mystery. With PlanetScale, Karl and his team were able to start with two shards and then re-shard as needed.
“Planning for massive scale while also rapidly iterating on our feature set was a demanding place to be—especially for a small team at a startup—and PlanetScale was able to deliver world class operations, a scalable platform, and the flexibility the business needed to grow both the member base and the feature set. It just does what we need.”
Of course, all of this is just icing on the cake. Ultimately, it came down to speed of feature development for the Community engineering team. Previously, feature changes often required rewriting and touching the data. When asked about how much time they now spend on managing their database the response was “honestly, nobody works on it most of the time.” The ability to arbitrarily add fields to their data model, and to join against other tables, meant they no longer had to worry about making their data fit their growing business.