PlanetScale added a new capability to its serverless database-as-a-service platform that aims to accelerate queries with an integrated form of data caching. PlanetScale Boost is currently in beta, with general availability expected to come sometime in 2023.
PlanetScale is based on the open source Vitess project, which provides a cloud-native database technology. The PlanetScale DBaaS launched in 2021 and has received a steady stream of updates in 2022, including a schema migration service that launched in March and an insights service in June that provides visibility into database operations.
The new PlanetScale Boost service uses the insights capability to identify the most used queries and then provides an integrated data caching functionality to accelerate those queries. The Boost technology, introduced Tuesday, uses the VStream change notification service in the Vitess open source project.
PlanetScale competes against a number of developer-focused databases, including FaunaDB and CockroachDB. With the addition of the Boost feature, PlanetScale will provide a data caching feature that could compete with other vendors in the data caching area, including Redis and Amazon's ElasticCache feature.
What makes PlanetScale Boost different than just a typical data cache is that the PlanetScale service allows for query acceleration without any additional application logic or new infrastructure, said Paul Nashawaty, an analyst at TechTarget's Enterprise Strategy Group.
Paul NashawatyAnalyst, ESG
"PlanetScale Boost has more functionality than traditional caching, as it simplifies the query patterns using insights but also can be deployed rapidly," Nashawaty said. "But this does not require additional caching infrastructure to maintain."
How PlanetScale boost works as a new data caching technology
Boost is not a typical data cache. Rather, it's based on academic research that describes a streaming data flow model for accelerating queries.
Typically with a data query, a series of operations happen in a database to provide what is known as a materialized view to deliver the result the user is looking for. The process of accessing a database and getting the correct result takes time and computing cycles.
At its core, PlanetScale Boost is a query materialization engine, said Nick Van Wiggeren, the vendor's vice president of engineering.
"A database administrator can pick a query that is running slow, add it to PlanetScale Boost, and we will actually launch a cache sitting right next to the database that will materialize the results of that query," Van Wiggeren said. "It will also keep the results of that query up to date."
Van Wiggeren said PlanetScale has seen queries that would normally take 500 millisecond to generate a response get a response in only 5 milliseconds in testing.
Bringing insights to data caching
Typically, an external data cache for a database requires customization and tuning to make sure the right queries are being cached and kept properly updated, as well as handling any underlying database schema changes.
With Boost, PlanetScale has integrated the service so the management and configuration is closely aligned with the database. With the PlanetScale insights service that launched in June, users of the DBaaS can identify which queries are taking longer. Users can pick whatever queries they want from the live database to be boosted.
With some types of data caching scenarios, an organization will deploy a cache in a different region than the primary database in order to accelerate a query for users within a certain geography. That's not how PlanetScale Boost works, however.
Instead, the data caching that Boost provides is deployed next to the primary PlanetScale deployment that it is supporting.
PlanetScale has not provided a public pricing for the Boost feature, though Van Wiggeren said the vendor's goal is for it to be less expensive for an organization than needing to scale up database availability to add more capacity. With Boost deployed, an organization will require fewer database resources to deliver the same volume of queries, he said.
"We're giving people basically a way to make their database do less work," Van Wiggeren said.