Serverless database: cheap, flexible and scalable-for the right application

MongoDB and DataStax recently released a serverless version of their cloud database. This points to the evolutionary direction of the database.

For many developers, serverless cloud computing in the form of AWS Lambda, Azure Functions will be part of the furniture, providing on-demand infrastructure with minimal configuration, scaling as needed, and consumption. You will be charged according to.

Lesser known are serverless databases such as DataStax AstraDB, which went public in February, and MongoDB Atlas Serverless, which was released last month. As with serverless computing, it also promises infinite scalability and consumption-based pricing in a way that’s especially suitable for cloud-native and event-driven workloads.

Serverless databases are designed to handle unpredictable and rapidly changing tasks. Are they really “serverless”?No, but not Lambda et al.. The server is there and humming in a data center somewhere, but you don’t have to worry about the configuration of the server and its virtualized descendants. Turn it on like a tap and turn it off again when you’re done. It’s an idea anyway. “Serverless” is one of the most deceptive terms in the industry, but the nuisance of linguistic purists is at a standstill.

Not convinced that serverless is an evolution is Akira Kurokin of MongoDB, the product owner of Percona.

“Why is it serverless when it’s server-full?” He may find it cheaper to run that way than per-server for small cloud databases that are used intermittently, such as in data science experiments. I said before admitting that. “This depends on how MongoDB will price in the future.”

Currently still in preview mode, MongoDB Atlas Serverless costs $ 0.30 per million dollars for reads, $ 1.25 per million dollars for writes and storage, and $ 0.25 per GB per month.

DataStax AstraDB, a serverless Cassandra service, is also offered at a similar price, and pricing will vary slightly depending on the choice of cloud provider. On EMEA, reads are $ 0.39, $ 0.26, and $ 0.26 on GCP, AWS, and Azure, respectively, and writes prices range from $ 1.16 to $ 1.33. The storage cost is $ 0.25GB / month.

Nanna Einarsdóttir, Vice President of Engineering for the Icelandic company Ankeri, a customer of DataStax AstraDB, said: Computing About the appeal of the company’s model, which provides real-time data to the commercial shipping sector.

First, the company said it had to handle large amounts of fast-moving data, so it decided on NoSQL as its primary data store architecture, and eventually surged Apache Cassandra for its scalability. After that, her team decided that the serverless option was suitable because there was no administrative overhead.

Estimate capacity in advance, no additional charges, just pay for the services you use over time-Nanna Einars dóttir, Ankeri

“We chose serverless Cassandra because we can run it as a service without having to implement our own clusters or perform administrative tasks,” says Einarsdóttir.

“Our developers can use APIs to work with data and the service handles all administrative tasks.”

“You don’t have to estimate capacity in advance and pay extra, you just pay for the services you use over time,” she added.

So far, the results have been in line with expectations, and serverless has allowed the team to run “much faster than if they had to design and install their own cluster,” she said. explained.

Einars dóttir continues. “By making applications and data serverless together, we can develop operations faster.”

Cassandra seems to be more involved than the serverless model, but it inevitably has a learning curve.

“In our initial design, we planned to include many columns in our database implementation to capture all the data provided by each ship, but this impacts performance each time a search operation is performed. I could give it, “she added. Engineers were waiting to help with a more compact and storage-efficient design.

MongoDB couldn’t provide an interview to the customer because Atlas Serverless is still in preview, but the benefits are about the same: cost control, scalability, rapid developer iteration, and reduced management. “MongoDB Atlas serverless instances provide seamless deployment and scaling, a reliable back-end infrastructure, and an intuitive pricing model,” said the company’s website.

DataStax and MongoDB are not the only database providers with serverless options. Leading cloud companies have been offering their own services for several years, including AWS Aurora Serverless, Azure SQL Serverless, and Google Firebase, and ISVs such as CockroachDB are also creating serverless alternatives.

reference: Large storage: Choose the right option for Kubernetes

Patrick McFadin, Vice President of Developer Relations at DataStax, believes that serverless databases that can run in multiple clouds will provide developers with new and flexible options.

“We had to rethink our internal architecture, but we were ready for this type of deployment based on how it runs as a distributed database,” he said of AstraDB, saying that changes would flow into the community version of Cassandra. I added.

“Share these changes with the open source community. This is a big change in the database. Which open source database will run serverless?”

Due to the temporary growth of cloud infrastructure, it can be wasteful to keep database connections on all the time. In cloud-native scenarios (or scenarios where applications take advantage of many serverless features), a database that connects only when needed and can grow automatically and almost instantly seems to be much better. .. It’s not surprising to see vendors moving in this direction, as there are other storage options available to cloud-native developers as well. Serverless can also be favored by small teams that lack the ability to manage their infrastructure.

Nevertheless, they are not suitable for all applications. There is an inevitable delay when the connection is established, and a cold start means an increase in delay. Teams may also find that the time saved in setup is being exhausted by the need to monitor resource usage and performance and implement new security measures.

That said, more organizations are expected to investigate serverless options in the future.

Serverless database: cheap, flexible and scalable-for the right application

Source link Serverless database: cheap, flexible and scalable-for the right application

Back to top button