Azure Cosmos DB serverless now in preview

This post has been republished via RSS; it originally appeared at: Microsoft Developer Blogs.

This blog post was co-authored by Nikisha Reyes-Grange, Senior Product Marketing Manager, Azure Marketing. We are beyond excited to announce that Azure Cosmos DB serverless is now available in preview on the Core (SQL) API, with support for the APIs for MongoDB, Gremlin (graph), Table, and Cassandra coming soon. This new consumption-based model lets you use your containers cost-effectively, without having to provision any throughput.

What is serverless on Azure Cosmos DB?

Announced at the Build conference this past May, serverless radically changes the way you use and pay for your Azure Cosmos DB resources. Until now, your only option for running a workload on Azure Cosmos DB has been to provision throughput. This throughput, expressed in Request Units (RUs) per second, is dedicated to your usage and available for your database operations to consume. This model comes with low-latency and high-availability SLAs, making it a great fit for high-traffic applications needing guaranteed performance - but not as well-suited to light traffic scenarios. Serverless is a cost-effective option for databases with sporadic traffic patterns and modest bursts. When your resources sit idle most of the time, it doesn’t make sense to provision and pay for unneeded per-second capacity. As a consumption-based option, serverless eliminates the concept of provisioned throughput and instead charges you for the RUs your database operations consume. Check this video for a visual comparison between provisioned throughput and serverless:

Differences with provisioned throughput

Serverless containers expose the same capabilities as containers created in provisioned throughput mode. This means that you manage your resources and read, write and query your data the exact same way. Where serverless accounts and containers differ from those using provisioned throughput is:

	Provisioned throughput	Serverless
Maximum number of Azure regions per account	unlimited	1
Maximum throughput burstability per container	unlimited	5,000 RU/s
Maximum storage per container	unlimited	50 GB

Note: Some of these limitations may be eased or removed as serverless becomes generally available, and your feedback will help us decide! Please reach out and tell us more about your serverless experience. When generally available, serverless will also offer different performance characteristics as explained in this article.

Use-cases

Because serverless is much more cost-effective in situations where you deal with light traffic and don’t need provisioned capacity, it naturally fits the following use-cases:

• Development
• Testing
• Prototyping
• Proofs of concept
• Non-critical applications with sparse traffic

To illustrate the advantages of using serverless for these scenarios, let’s take the example of a workload that is expected to burst to a maximum of 500 RU/s and consume a total of 20,000,000 RUs over a month.

• In provisioned throughput mode, you would have to provision a container with 500 RU/s for a monthly cost of: $0.008 * 5 * 730 = $29.20
• In serverless mode, you would only pay for the consumed RUs: $0.25 * 20 = $5.00

(not accounting for the storage cost, which is the same between the 2 modes)

Getting started

You must create a new Azure Cosmos DB account from the Azure portal to get started with serverless. When creating your new account, select Core (SQL) as the API type, then Serverless (preview) under Capacity mode: Once your new serverless account is created, you manage your Azure Cosmos DB resources and data just like you would in provisioned throughput mode. The only difference is that you don’t have to specify any throughput when creating a container: We can’t wait to see what you will build on Azure Cosmos DB serverless! Get started today with the following resources:

• Serverless documentation
• Azure Cosmos DB pricing page
• How to choose between provisioned throughput and serverless