MongoDB Atlas Introduces Enhanced Cost Optimization Tools

MongoDB Atlas was designed with elasticity at its core and has always allowed customers to scale capacity vertically and horizontally, as required and automatically. Today, these inherent capabilities are even better and more cost-effective. At the recent MongoDB.local London, MongoDB announced several new MongoDB Atlas features that improve elasticity and help optimize costs while maintaining the performance and availability that business-critical applications demand. These include scaling each shard independently, extending storage beyond 4 TB or more, and 5X more responsive auto-scaling.

Organizations and their customers are inherently dynamic, with operations, web traffic, and application usage growing unpredictably and non-linearly. For example, website traffic can spike due to a single video going viral on social media, and holidays are a frequent cause of application usage slowdowns.

Traditionally, organizations have tackled this volatility by over-provisioning infrastructure, often at significant cost.

Cloud adoption has improved the speed at which infrastructure can be provisioned in response to growing and volatile demand. Simultaneously, companies are focused on striking the perfect balance between performance and cost efficiency. This balance is acute in the current economic climate, where cost optimization is a top priority for Infrastructure & IT Operations (I&O) leaders.

However, scaling infrastructure to meet demand without overprovisioning can be complex and costly. Organizations have often relied on manual processes (like scheduled scripts) or dedicated teams (like IT ops) to manage this challenge. MongoDB Atlas enables a more effective approach. With MongoDB Atlas, customers can manage flexible provisioning, zero-downtime scaling, and easy auto-scaling of their clusters. From October 2024, all Atlas customers with dedicated tier clusters can employ these recently announced enhancements for improved cost optimization.

Granular resource provisioning

MongoDB’s tens of thousands of customers have complex and diverse workloads with constantly changing requirements. Over time, workloads can grow unpredictably, requiring scaling up storage, compute, and IOPS independently and at differing granularities. Imagine a global retailer preparing for Cyber Monday, when traffic could be 512% higher than average— additional resources to serve customers are vital.

Independent shard scaling enables customers running MongoDB Atlas to do this in a cost-optimal manner. Customers can independently scale the tier of individual shards in a cluster when one or more shards experience disproportionately higher traffic. For customers running workloads on sharded clusters, scaling each shard independently of all other shards is now an option (for example, only the shards serving US traffic during Thanksgiving). Customers can scale operational and analytical nodes independently in a single shard.

This improves scalability and cost-optimization by providing fine-grained control to add resources to hot shards while maintaining the resources provisioned to other shards. All Atlas customers running dedicated clusters can use this feature through Terraform and the Admin API. Support for independent shard auto-scaling and configuration management via the Admin API and Terraform will be available in late 2024.

Extended Storage and IOPS in Azure: MongoDB is introducing the ability to provision additional storage and IOPS on Atlas clusters running on Azure. This enables support for optimal performance without over-provisioning. Customers can create new clusters on Azure to provision additional IOPS and extended storage with 4TB or more on larger clusters (M40+). This feature is being rolled out and will be available to all Atlas clusters by late 2024. Head over to our docs page to learn more.

With these updates, customers have greater flexibility and granularity in provisioning and scaling resources across their Atlas clusters on all three major cloud providers. Therefore, customers can optimize for performance and costs more effectively.

More responsive auto-scaling

Granular provisioning is excellent for optimizing costs while ensuring availability for an expected increase in traffic. However, what happens if a website gets 13X higher traffic or a surge in app interactions due to an unexpected social media post?

Several enhancements to the algorithms and infrastructure powering MongoDB’s auto-scaling capabilities were announced in October 2024 at .local London. Cumulatively, these improve the time taken to scale and the responsiveness of MongoDB’s auto-scaling engine. Customers running dynamic workloads, particularly those with sharper peaks, will see up to 5X improvement in responsiveness. Smarter scaling decisions by Atlas will ensure that resource provisioning is optimized while maintaining high performance. This capability is available on all Atlas clusters with auto-scaling turned on, and customers should experience the benefits immediately.

Industry-leading MongoDB Atlas customers like Conrad and Current use auto-scaling to automatically scale their compute capacity, storage capacity, or both without needing custom scripts, manual intervention, or third-party consulting services. Customers can set upper and lower tier limits, and Atlas will automatically scale their storage and tiers depending on their workload demands. This ensures clusters always have the optimal resources to maintain performance while optimizing costs. Take a look at how Coinbase is optimizing for both availability and cost in the volatile world of cryptocurrency with MongoDB Atlas’ help, or read our auto-scaling docs page to learn more.

Optimize price and performance with MongoDB Atlas

As businesses focus more on optimizing cloud infrastructure costs, the latest MongoDB Atlas enhancements— independent shard scaling, more responsive auto-scaling, and extended storage with IOPS—empower organizations to manage resources efficiently while maintaining top performance. These tools provide the flexibility and control needed to achieve cost-effective scalability.