ScyllaDB operates in the data infrastructure sector, with a NoSQL database intended for services that need to handle large numbers of simultaneous reads and writes. Based in the United States, the company targets digital platforms, e-commerce and interactive applications in particular. Its positioning: making efficient use of hardware resources to reduce response times and operating costs. Its website, scylladb.com, brings together documentation, commercial offerings and technical resources.
From virtualization specialists to databases
ScyllaDB was founded by Dor Laor and Avi Kivity, two engineers known in particular for their work on KVM, a virtualization technology integrated into Linux. Publicly introduced in 2015, their database adopts Apache Cassandra’s distributed model but relies on a different implementation, written in C++. The aim is to retain a familiar interface for teams while rethinking the use of processors, memory and input/output.
This approach relies on Seastar, an asynchronous programming framework. The architecture distributes work across processor cores while reducing resource sharing and locking mechanisms. This technical choice is one of the product’s main differentiating features.
A distributed database for intensive applications
ScyllaDB stores data using a wide-column model and distributes it across multiple nodes. Replication helps maintain service availability, while adding machines accommodates increasing workloads. Compatibility with CQL and Cassandra drivers makes some migrations easier, without eliminating the need to validate schemas, queries and application behavior.
Use cases include user profiles, catalogs, event histories and messaging. Discord is among the companies that have documented their use of ScyllaDB. Through Alternator, the product also offers an interface compatible with the Amazon DynamoDB API. This option broadens deployment scenarios but does not remove the need to check whether the required features are supported.
The company generates revenue from its software offerings, support and ScyllaDB Cloud, its managed service. This service handles some operational tasks, such as provisioning, backups and updates. The commercial proposition focuses as much on operational control as on performance, which depends on hardware, the data model and the actual workload.
What’s next?
For ScyllaDB, the challenge will be to turn its architectural choices into verifiable economic gains compared with services from major cloud providers and other distributed databases. Ease of migration, cost predictability and operational automation will influence purchasing decisions. Its progress will also depend on its ability to make its performance accessible without requiring excessive expertise from customer teams. In this market, tests with representative workloads remain more informative than theoretical comparisons.