DataStax has this week announced that its Database-as-a-Service product - Astra - will now be made available with a serverless architecture, which separates compute from storage, enabling database resources to scale up and down on demand to match application requirements.
Astra is based on the distributed Apache Cassandra open source database, and when released early last year, was a clear signal to the market that DataStax is doubling down on ‘ease of use' for developers in the cloud.
Following the launch of Astra, DataStax swiftly followed with the release of Stargate, a data gateway that allows developers to use any data store for apps, by adding support for new APIs, data types and access methods. Again, a clear focus on ease of use and flexibility for developers.
And whilst today's serverless announcement is being hailed as a database ‘industry first', and will likely excite many developers using Cassandra, it should be viewed within the context of DataStax shifting the view that its tools are more difficult to work with than some of its competitors (for example, MongoDB).
Whilst Cassandra has a solid reputation for being highly scalable and solidly underpinning ‘applications of the internet', in the past it's not something that developers would typically have chosen for all use cases. DataStax is hoping that its recent announcements - in particular today's serverless one - go some way to changing that.
DataStax claims that the benefits of Astra as a serverless DBaaS can "bend the curve" of escalating data costs, delivering enterprises total cost of ownership savings of up to 3 to 5 times over non-serverless database workloads.
We got the chance to speak with Ed Anuff, chief product officer for DataStax, ahead of today's announcement - both about the serverless architecture itself and its application in the enterprise, but also DataStax's ambitions. Speaking to the benefits of going serverless, Anuff said that there are gains to be made for both developers and the business, as well as cost savings.
Firstly, he said:
From an enterprise IT standpoint, serverless is just way more efficient. Your total cost of ownership is significantly cheaper. Typically with data you're sizing to your peaks. Obviously now with a serverless capability, you don't worry about that, so your overall cost becomes a lot less.
In addition to the cost aspect, developers will now find it easier to make Cassandra their first choice. The process of selecting a database for a specific use case is time consuming and ultimately costly, Anuff said. Serverless removes a lot of these early problems in the development cycle. He said:
In terms of the specific thing you can do with it, if you're a developer and starting a project, you can just go and say I'm going to start with Cassandra from the outset. What developers typically say is that Cassandra gives scale, but running these large clusters is expensive and complicated, so they use it when they need to use it.
So by giving them this serverless capability, you don't have to get into that question of ‘do I really need to use it?'. It's easier to just go and make Cassandra your first choice and gets past all the concerns that take a whole bunch of time.
In addition to the above benefits, serverless ultimately gives the business more flexibility to test and scale new initiatives, without having to worry about capacity planning. Anuff said:
What you now have is the scale when you need it, which makes a lot of different types of business programmes much more feasible. If somebody comes in and says they want to go and do some promotion next week, or launch a new feature that is database intensive, it now makes it possible for you to do this. So you get developer agility, business agility, and the costs are much more reasonable.
Anuff said that if an enterprise's workloads are highly variable - say, for example, retailers that build out capacity for specific peaks - then the cost savings to the business can be "profound". And whilst DataStax is still continuing to offer its customers the option to go in and prescale their databases if they choose, the company has found in the Beta period that customers are typically starting with serverless first and foremost.
There aren't that many concerns that we see people raising. We think generally people should be thinking about how they transition to serverless. If you're selecting a NoSQL database it's something you should be thinking about - is there a serverless option available? For new projects you should just be starting on this.
Tangible change at DataStax
As noted above, DataStax has been on a journey over the past year or so to change its reputation amongst developers as a tool that is as easy to use as its competitors - which is being guided under new leadership, after former Google VP Chet Kapoor was appointed as CEO.
The combination of Astra, Stargate and now serverless, this positions DataStax as a strong contender for developers looking to use an agile, multi-cloud, scalable database, Anuff said:
If you look at what we've done and announced over the last year there should be a fairly consistent pattern to it. We've very systematically gone and said, what are the reasons why you are not using Cassandra? And then we have tried to address those things.
People would say, it's difficult to run. So we put it in the cloud and put it on Kubernetes. So now it's very easy to run. People said that it's difficult to develop for and they didn't like the data model and preferred some of the other NoSQL databases. So we went and introduced Stargate that makes it possible to use Cassandra with APIs, GraphQL, REST and JSON data - and we are adding more and more stuff to that.
And then finally people said that they've been using Cassandra, which gives them lots of power, but it's costly and they want it to be as efficient as possible. So now we have introduced serverless. So really overall our goal is to make Cassandra one of the first choices and one of the best choices for people that want to use a NoSQL database.