Expanded Memory and Compute with Heroku's New Larger Dynos

Introduction

Heroku is excited to introduce nine new dyno types to our fleets and product offerings. In 2014, we introduced Performance-tier dynos, giving our customers fully dedicated resources to run their most compute-intensive workloads. Now in 2024, today's standards are rapidly increasing as complex applications and growing data volumes consume more memory and carry heavier CPU loads.

With these additional dyno types, we’re excited to enable new use cases on Heroku with enhanced compute and memory specifications. Some use case examples include real-time processing against big data/real-time analytics, large in-memory cache applications such as Apache Spark or Hadoop processing, online gaming, machine learning, video encoding, distributed analytics, and complex or large simulations.

Heroku is addressing these modern developer requirements with three new dyno types for each of our Performance, Private, and Shield dyno tiers:

• Performance-L-RAM, Performance-XL, and Performance-2XL for Heroku Common Runtime
• Private-L-RAM, Private-XL, and Private-2XL for Private Spaces
• Shield-L-RAM, Shield-XL, and Shield-2XL for Shield Private Spaces

What’s New

We created three distinct new dyno sizes for each of the Performance, Private and Shield tiers that allow for increased flexibility and higher performance ceilings for Heroku customers.

• Performance/Private/Shield-L-RAM: If you’re targeting more memory-focused tasks, you can lower your CPUs, and double your RAM while maintaining the same cost as existing *-L dynos. These dynos are perfect for tackling memory-intensive tasks like large-scale image processing or data analysis.
• Performance/Private/Shield-XL: quadruples your RAM while providing the same stellar CPU performance as *-L dynos, empowering you to run simulations or deliver lightning-fast processing.
• Performance/Private/Shield-2XL: delivers a staggering 8x RAM and 2x CPU boost compared to our *-L dynos, unleashing the full potential of your ambitious projects.

See the updated dyno table for how these new dynos stack up to our previous offering:

You can migrate applications in seconds using simple CLI commands or through the Heroku Dashboard.

Pricing information is transparent and costs are prorated to the second, so you only pay for what you use. Visit the Heroku pricing page for more details and the Heroku Dev Center on how to unlock more power with these new dynos.

Get Started with New Dynos

All Heroku customers interested in using our new Performance dynos for their applications can start today. The process is simple and follows the typical process of spinning up and switching dyno types.

To provision these dyno types from the Heroku Dashboard, follow the Heroku Dev Center steps on setting dyno types.

Or simply run the following CLI command:

$ heroku dyno:type performance-2xl 

Private Space customers can also use the new Private Dynos, and Shield Private Space customers can use the new Shield Dynos in their spaces.

How Heroku Uses the New Dynos

As we started to internally test and prepare the new dyno types for general availability, the Heroku Connect team was a prime candidate as an internal customer. Its data-intensive operations power the Heroku Connect product offering, which enables developers seamlessly access Salesforce CRM data using Heroku Postgres. This bi-directional data synchronization requires hundreds of Shield dynos to make sure data is up-to-date and accurate between Salesforce and Postgres. With a growing number of Heroku Connect customers, the Connect team was reaching the memory limits of our Shield-L dynos, requiring constant scale-ups to meet customer demands.

At the beginning of February, the Connect team upgraded their dyno fleets from Shield-L to Shield-XL dynos. After monitoring the platform and re-scaling appropriately, the team successfully reduced the total number of dynos required to run the data synchronization. The new formation continued to meet all of the availability and data quality requirements that Connect customers expect. In total, by changing their formation to utilize the new dyno sizes, the team reduced the estimated compute-specific costs of running Heroku Connect jobs by almost 20%!

From a senior engineer on the Heroku Connect team:

"We were able to reduce cost and reduce the number of dynos we needed because a lot of these operations are memory-heavy. With the newer dynos, we overcame this bottleneck of memory which required us to add more dynos in the past".

We hope that our customers can perform the same cost optimizations unlocked by these new dyno offerings. This launch is another step towards making Heroku a more cost-effective cloud platform.

We’re excited for the internal wins for our Heroku teams. We’re even more excited to see what new projects and optimizations are possible for our customers now that these dynos are generally available.

Conclusion

With the new larger dyno types, we’re pushing the boundaries of what is possible with Heroku. We’re working to make our platform, bigger, faster, and more resilient. We’re continuously listening to our customers on our Github Public Roadmap. The valuable feedback on the Larger Dynos roadmap item led to this change.

Paired with our recently announced plans for flexible storage on Heroku Postgres, we're working hard to make sure Heroku can scale with your business.