DataStax, the company behind the Cassandra database, recently announced the general availability of Astra Serverless, the open, multi-cloud serverless database-as-a-service (DBaaS) solution.
DataStax Astra builds upon Apache Cassandra database and introduces a microservices-based architecture that separates compute from data storage tasks. Database resources can scale up and down on demand to match application requirements and traffic independent of compute resources.
While serverless compute has been around for a while, serverless data has lagged due to the technical challenges around separating compute and storage. Serverless compute is about stateless serverless workloads like web apps and functions, often running on Kubernetes, that delegate persistent data storage to other systems. And serverless data focuses on stateful processes (caches, databases, data stores, state machines, etc) that read, write, persist, and transmit data. These have been challenging workloads to run on Kubernetes. Scaling a database typically requires the addition of more server nodes in order to handle more demand or to store more data, which requires that the entire data set be “rebalanced” across the nodes to keep the ratio of storage and computing capability equal.
With Astra serverless, developers will only pay for what they use, no matter how many database clusters they create and deploy. This lets developers and IT teams create the databases they need for development, testing, staging, or other purposes.
The Astra Serverless database service scales up and down with a customer's application requirements and the traffic. It's recommended for the following use cases:
- Applications with unpredictable traffic and data consumption patterns. Developers can create an end-point and let the serverless database auto-scale per changing requirements of their applications.
- Applications with variable workloads depending on the season or timing of data processing like HR, Budgeting, Tax filing etc.
- Non-production databases like Dev/Test databases that are used only during the working hours of the organization. Serverless scales down to zero automatically when not in use.
Last year DataStax introduced the Astra DBaaS to make Cassandra available as a cloud-based database, and the Stargate API Gateway to help build applications on Cassandra using JSON, REST, and GraphQL APIs.
Serverless data APIs are becoming popular in bringing the different types of databases to the cloud, with offerings like Amazon's Aurora Serverless Data API.
InfoQ spoke with Pieter Humphrey, developer product manager at DataStax, about the new announcement and how the serverless Astra solution helps the developer community.
InfoQ: Why is serverless technology a better choice for database-as-a-service architecture?
Pieter Humphrey: Dynamic auto-scaling based on workload demand can result in tremendous cost savings, just as we've seen with PaaS vs. Serverless workloads.
What can people do with those savings? Lots! Some examples: lower costs to run in production since it scales to zero when not used. Developers no longer need to share development environments, as it's economical enough for each developer to have their own. Automated test suites can be done more frequently, or additional tests can be added, implying longer individual test suite runs, improving code quality. Sophisticated auto-scaling can eliminate an entire class of DIY IaC (infra-as-code) code or CM (config management) tool work with things like Puppet, Chef, Ansible, Terraform, etc.
InfoQ: How does Stargate data API Gateway work with the new Astra serverless DBaaS?
Humphrey: DataStax Astra is delivered as a serverless database-as-a-service that comes with the Stargate data API gateway.
Stargate works the same way it did before the introduction of serverless Astra. Since Stargate is what segregates Cassandra nodes into two different roles - query coordination (compute) and storage (data r/w) - it actually could be seen as a precursor for serverless Astra.
InfoQ: What does the future roadmap for serverless Astra look like?
Humphrey: We believe serverless Astra is a game changer for the industry, and we will continue to quickly advance it to benefit developers and enterprises. Now that running a database isn't tied to running fixed infrastructure, you can expect more Cassandra integrations with serverless technologies like Serverless Framework, Lambda, OpenFaaS and provisioning services like Terraform and Jenkins.
DataStax is committed to open source and to supporting the projects we participate in by leading with code. As a part of that commitment, we will be donating the changes we make in Apache Cassandra back to the project to better enable a cloud native future for Cassandra. We hope to see enhancements from k8ssandra, Stargate, ad Astra's serverless architecture improvements in the Cassandra project after 4.0 GA, along with innovations from other members of our community.
If you are interested in learning more about the Astra serverless product and how a serverless data API helps with DBaaS initiatives, check out the upcoming webinar.