AWS re:Invent 2025 - What's new with Amazon EBS (STG202)

AWS re:Invent 2025 - What's New with Amazon EBS

Overview of Amazon EBS

• EBS (Elastic Block Store) is a network-attached persistent storage for EC2 instances
• EBS offers different volume types (IO2, GP3) for various workload requirements
• EBS provides the ability to snapshot data, which is backed by Amazon S3 for high durability
• EBS is a distributed system maintained and operated by AWS engineers to offer performance, durability, and features

Real-Time Transactional Systems

Performance

• Mission-critical applications require consistent sub-millisecond latency to deliver a seamless customer experience
• EBS recommends IO2 volumes for these workloads, with a tightened latency guidance of sub-500 microseconds
• Customers can understand the difference in latency between IO2 and general-purpose volumes to make the right price-performance trade-off

Global Infrastructure Parity

• As customers expand globally, they need the same infrastructure available across regions to reduce operational burden and bring data closer to customers
• EBS has expanded IO2 Block Express volumes to all commercial and GovCloud regions to meet this need

High Resiliency

• Observability is crucial for customers to understand performance and identify bottlenecks
• EBS has launched metrics in CloudWatch for average latency, IOPS, and throughput, as well as "exceed checks" to signal when performance thresholds are reached
• Chaos testing is essential for building resilient systems, and EBS has introduced the ability to inject latency faults using the Fault Injection Simulator (FIS) service
• EBS Recycle Bin provides data protection by allowing recovery of accidentally deleted volumes within a specified retention period

Analytical Warehouse Systems

• Analytical warehouses require high-throughput and large-capacity storage to handle the influx of data from various sources and run complex queries
• EBS has introduced GP3 volumes with 5x higher IOPS (80K), 2x higher throughput (2GB/s), and 4x larger size (64TB) to meet these demands
• Larger GP3 volumes simplify test/dev workflows by allowing the same volume size as production IO2 volumes
• Increased network bandwidth (up to 150Gbps) from R8 instances ensures the compute can fully utilize the high-performance EBS volumes

Developer Environments

Volume Cloning

• EBS Volume Clones enable instant creation of a new volume from an existing one, without the need for snapshots as an intermediate step
• This improves developer efficiency by allowing more frequent copies of production environments for testing, and enables simpler blue-green deployments

Golden Image Workflows

• Customers create "golden" Amazon Machine Images (AMIs) to provision consistent development environments across regions
• EBS introduced time-based Army (AMI) copying to ensure predictable and fast deployment of these images across regions
• Provision rate for volume initialization ensures developers have a fully performant volume ready to use, without waiting for the data to be pulled from snapshots

Key Takeaways

• EBS has made significant improvements to address the needs of mission-critical transactional systems, analytical warehouses, and developer environments
• Performance, global availability, and resiliency are enhanced through features like tighter latency guarantees, expanded IO2 volumes, observability metrics, and chaos testing
• Larger and faster GP3 volumes simplify analytical and test/dev workflows
• Volume cloning and golden image workflows improve developer agility and productivity
• EBS continues to evolve as a robust, reliable, and high-performing storage service for a wide range of workloads on AWS