kentoh - Fotolia
DataStax said on Thursday that its Astra cloud database serverless platform will be generally available on March 4.
The new serverless platform is an update to the original DataStax Astra platform that is based on the open source Apache Cassandra database; Astra first became available in May 2020.
Until now, DataStax Astra had been available in a model for which users needed to choose an appropriate sizing for their database requirements. That sizing included configuring a certain amount of compute and storage resources. With the serverless approach, DataStax Astra will now expand and contract resources as needed, without users needing to define a specific size.
The idea of a serverless database is an attractive ideal that should appeal to many organizations, said Carl Olofson, an IDC analyst.
"I think it will grow in importance as people become more sensitive to the nuances of the pay-as you-go or pay-as-you-grow model when, in many cases, bills go up, but don't come down," Olofson said. "I think serverless will be a growing trend in the database industry."
Existing DataStax Astra users can choose serverless
DataStax already has paying customers for its original Astra cloud database platform, which is not serverless. Ed Anuff, chief product officer at DataStax, explained that moving forward with the serverless model, all new users and their databases will be provisioned with the serverless approach.
Anuff noted that for users of the existing DataStax Astra platform, which he referred to as the "classic" platform, there will be an option to upgrade their database to the serverless model.
Carl OlofsonAnalyst, IDC
"From an operations standpoint, you don't want to make any changes underneath somebody without them being aware of it," Anuff said. "But we do expect that most users will want to do upgrade as the cost savings are pretty significant."
How DataStax Astra serverless works
With the classic edition, as well as with how the open source Cassandra database works in general, users need to define a certain amount of compute and storage for a deployment.
As demand grows with Cassandra, a user could add capacity to scale out the database. The real challenge, however, was that Cassandra did not have the ability to shrink resources when demands contracted. One such use case in which demand fluctuates is retail, which uses a lot of Cassandra.
With "retailers that use Cassandra, they would have to figure out what their peak demand looks like, which might typically be over the holidays," Anuff said. "So that would mean that in most cases, if you're sizing for your peak capacity, that 90% of the time, you actually have much more capacity than you need."
The DataStax Astra serverless approach changes that situation, so that users only pay for what they need.
Kubernetes helps enable the DataStax serverless model
DataStax has been busy in recent months working to optimize Cassandra deployment and operations with the open source Kubernetes container orchestration system, including developing an enhanced Kubernetes Operator.
Kubernetes allows for compute resources to be scaled up and down as needed. That's a core part of the serverless platform, but it's not the only piece, Anuff said.
A key part of enabling serverless is separating compute and storage resources, he said.
"What we have done is taken Cassandra and we've reengineered its storage tier so its separated from the compute nodes," Anuff said. "The work that we've done to adapt Cassandra for Kubernetes was a big part of what made serverless possible, but also this idea of separating storage is what makes it possible for us to leverage it."