ScyllaDB updated its namesake database with a series of capabilities that aim to improve the open source NoSQL database.
ScyllaDB V became generally available on July 12.
During the development of the new release, the vendor had previously referred to the release as ScyllaDB 5.0, but has rebranded it to Roman numeral V for marketing purposes.
With ScyllaDB V, the database gets a new scheduler that helps to optimize read/write operations. The database also has a new model for user-defined functions (UDFs) that uses WebAssembly technology that can support multiple programming languages, including C, JavaScript, Python and Rust.
ScyllaDB competes against multiple database vendors, including Apache Cassandra-based DataStax and Amazon with its DynamoDB system. Unlike DataStax or Amazon, ScyllaDB is not yet offering a database-as-a-service model, though it plans on updating its cloud service later this year to a consumption-based approach.
Among ScyllaDB's users is online advertising technology vendor GumGum, which uses the NoSQL database to tally and count views for ads.
Keith Sader, director of engineering for ad server at GumGum, said he's interested in the ScyllaDB V performance improvements that will enable better management of workload scaling.
Sader said he's also looking forward to ScyllaDB's rollout of a consumption-based database-as-a-service model. Currently, ScyllaDB Cloud runs as a service inside an organization's own virtual private cloud environment in AWS.
Over time, we noticed that sometimes the latency that we hoped to get isn't what we were getting in certain cases, and we learned that storage can perform differently for different read and write workloads.
Dor LaorCEO, ScyllaDB
"We would be interested in purchasing our ScyllaDB usage like we purchase every other system, on AWS as a pay-as-you-consume basis versus everything upfront," Sader said. "This would help our DevOps folks focus on the core business issues of our platform versus managing infrastructure."
ScyllaDB V accelerates NoSQL database operations
At the core of the ScyllaDB V update are performance upgrades designed to improve database consistency and accelerate workloads.
Among the new features is an I/O scheduler. ScyllaDB's previous I/O scheduler took a "simplified" view of storage performance, understanding the throughput and latency for access, said Dor Laor, CEO of ScyllaDB.
"Over time, we noticed that sometimes the latency that we hoped to get isn't what we were getting in certain cases, and we learned that storage can perform differently for different read and write workloads," Laor said.
The new I/O scheduler accounts for the characteristics of different workloads and can understand how storage access for read/write will perform. The updated scheduler also enables users to scale a ScyllaDB cluster with lower latency.
ScyllaDB brings WebAssembly support to NoSQL database operations
All databases provide an integrated set of functions. When an organization needs a custom function, UDFs come into play by enabling an organization to write its own custom code.
With ScyllaDB V, the database now uses WebAssembly for UDF, making it easier for developers to extend the database. With WebAssembly, an organization can write code in a number of different languages, without having to write in just one specific language that is supported by the database.
ScyllaDB supports the Cassandra Query Language syntax for creating UDFs. Before ScyllaDB V, though, the database only ran code written in the Lua programming language and not Java, which is the language used for UDFs in Apache Cassandra.
"Now, with WebAssembly, we can run any code for a user-defined function," Laor said.
