A New Era for Managed Databases: Announcing Heroku Postgres Advanced

At Heroku, we’ve built a platform that allows developers to focus on their apps, not managing their infrastructure. Heroku Postgres Advanced carries forward that philosophy as our next-generation database offering, available today.

Modern applications demand more from their databases. They need independent control over compute and storage, so you can scale each dimension based on your workload. They need dedicated resources for analytics workloads that won’t slow down your primary app. They need to scale smoothly with higher throughput.

Heroku Postgres Advanced delivers all of this. Built on entirely new, high-performance infrastructure, our new scalable managed database tier is ready for all workloads, from everyday applications to your most demanding production workloads.

A new architecture for flexible scalability and performance

Heroku Postgres Advanced is built on entirely new infrastructure that fundamentally changes how your database scales, performs, and handles different workloads. This architecture decouples compute from storage, introduces instance pools for workload isolation, and delivers higher throughput compared to the Postgres Classic tier. The result is a managed database that adapts to your application’s needs without app downtime or operational overhead.

Heroku Postgres Shield-10 eliminated our operational headaches that come during seasonal enrollment surges. As HealthSherpa scales to serve even more healthcare consumers, adopting Heroku Postgres Advanced will provide the performance, compliance, and flexibility we need for the next level of growth

Scale compute and storage independently

Match your resources to your app’s needs. No more overprovisioning compute just to get the storage you need. If your app is storage heavy but compute light, you can now scale to over 200 TB of storage without paying for additional compute you don’t need. When high-traffic spikes hit, you can scale your compute power vertically and horizontally without an app restart or committing to unused storage.

The decoupled architecture gives you precise control over your database resources based on your actual workload, not predefined tier plans.

Unprecedented throughput on entirely new infrastructure

Heroku partnered closely with AWS to power Heroku Postgres Advanced using their latest high-performance database infrastructure. In our internal benchmarks and pilot testing, we observed 3x to 4x higher throughput compared to similar Postgres Classic configurations.

From everyday databases to business-critical systems, these improvements apply to all your workloads. Applications that handle routine tasks, push heavy transaction volume, or run analytical queries, all see lower latency and higher overall performance.

This added efficiency lets you optimize costs by tailoring the configuration to what your application needs. To find the ideal setup, we recommend benchmarking performance using simulated real-world loads.

High-performance storage with predictable costs

Scale your storage without worrying about unpredictable bills or performance bottlenecks. Heroku Postgres Advanced introduces elastic storage built for high-concurrency workloads. While many database providers only allow storage to expand, our storage scales bidirectionally, automatically expanding or contracting based on your actual data footprint. You only pay for the database storage you use.

You don’t need to provision or manage IOPS (Input/Output Operations Per Second). High-speed I/O is built directly into the platform. This delivers consistent, low-latency performance during traffic spikes without the variable I/O costs found in traditional cloud databases.

Instance pools for workload isolation

Heroku Postgres Advanced introduces instance pools, allowing you to create a leader pool (read/write) and up to five independent follower pools (read-only).

• Horizontal scaling: Add up to 13 follower instances to a single cluster to distribute read load.
• Workload isolation: Name your pools (e.g., analytics-pool) and attach them to specific apps. Keep your leader pool fast while your BI tools crunch numbers on a dedicated follower pool, all reading from the same storage layer. Each pool has one endpoint, and traffic is loadbalanced across instances.
• Aggregate connections: Connection limits scale with your pool size. A pool of three instances, each allowing 5,000 connections, provides a total pool capacity of 15,000 connections.

Non-disruptive operations

Scaling, enabling High Availability (HA), and managing credentials now happen in the background. With our new interactive CLI commands, you can modify your database architecture on the fly with zero app downtime.

Thank you to our pilot participants

To the customers who participated in our pilot program and Limited GA: thank you. Your feedback on everything from the interactive CLI prompts to storage quota enforcement has shaped this launch. We’re also grateful for the continued feedback from our Postgres Classic users as we build the future of Heroku Data.

What’s next for Heroku Postgres

Getting started with Heroku Postgres Advanced

Ready to get started? Check out the Getting Started Guide to provision your first Postgres Advanced cluster today. For a deeper dive, visit our Dev Center documentation.

Heroku Postgres Advanced roadmap

The decoupled architecture of Heroku Postgres Advanced allows us to iterate faster and deliver new capabilities. We’re committed to reaching full feature parity with Postgres Classic tier plans while pushing the boundaries of what a managed database can do.

Here are the top priorities on our roadmap:

• Comprehensive migration tooling: An end-to-end self-service migration framework featuring automated preflight compatibility checks, flexible streaming or full-load methods, auto-promotion, and guided interactive CLI workflows.
• Native logical replication for real-time streaming: Stream your data in real time to external data warehouses, analytics platforms, or other Postgres instances. Built-in logical replication support enables complex ETL workflows and synchronizations.
• Streaming data connectors support: Stream real-time database changes from Heroku Postgres Advanced directly into Apache Kafka on Heroku using fully managed Change Data Capture (CDC).
  Autoscaling: Intelligent autoscaling that detects traffic spikes and automatically adjusts your compute resources. Maintain performance during peak loads while reducing costs during quiet periods, without manual intervention.
• Global clusters: Deploy follower clusters in different regions than your primary region. Place data closer to your users around the world, reducing latency for read-only global traffic.
• Server-side connection pooling: Built-in connection pooling to manage tens of thousands of concurrent connections efficiently. This removes the complexity of configuring client-side connection poolers and eliminates connection exhaustion errors.
• Sleep mode: Compute instances automatically pause when they are not in use. This is ideal for development and staging environments where you want to reduce costs for non-production databases.

We’ll continue shipping the features you rely on today alongside new capabilities that expand what managed Postgres can do. Reach out and let us know what features are important to you and your workloads.