AWS for Industries

The Road to 180M GRefs/s: Sizing Epic on AWS with R8ib and Enhanced EBS

Amazon Web Services (AWS) announces the availability of the new Amazon Elastic Compute Cloud (Amazon EC2) r8ib.96xlarge instance for healthcare workloads. When combined with the latest Amazon Elastic Block Store (Amazon EBS) io2 Block Express advancements, this new instance enables Epic’s Chronicles Operational Database (ODB) to be sized up to 180 million global references per second (GRefs/s) within a horizontally scaled Enterprise Cache Protocol (ECP) architecture.

The r8ib.96xlarge combines the cost-efficiency of the r8i family with enhanced EBS I/O performance, delivering a 28% GRefs/s increase over the previous generation to support Epic customers. Building on the same custom Intel Xeon 6 processor platform introduced with r8i, the r8ib variant adds 6th generation AWS Nitro cards to deliver significantly higher storage I/O performance for horizontally scaled ECP deployments.

Technical specifications

Building on the custom Intel Xeon 6 processor platform built specifically for AWS and introduced with r8i, the r8ib variant incorporates 6th generation AWS Nitro cards to deliver significantly higher storage I/O performance for horizontally scaled ECP deployments. AWS Nitro System, a lightweight hypervisor that delivers improved compute and networking performance while enabling efficient, flexible cloud services with enhanced security, isolated multi-tenancy, private networking, and fast local storage.

r8ib.96xlarge (May 2026)

The r8ib EC2 instance family builds on the r8i architecture with enhanced EBS performance to deliver up to 200% higher EBS I/O performance compared to r8i for Epic workloads. Key advancements include:

  • Up to 384 vCPUs and 3TiB DDR5 memory with optimized memory subsystem delivering significantly higher bandwidth per core
  • Increase in EBS performance over r8i delivering up to 1,440K IOPs and 37.5 GB/s bandwidth throughput with sub-millisecond latency for critical Healthcare workloads

Since July 2021, AWS increased ECP scalability for Epic from 18M to 180M GRefs/s, a 10x improvement across eight advances. Recent progress accelerated significantly: capacity grew from 75M to 180M in just over two years (April 2024 to May 2026), with individual jumps of 30–40M arriving every 3–6 months. Symmetric multiprocessor (SMP) scalability also grew 2.5x over the same period, from 20M to 49M GRefs/s. This expanded capacity means the largest Epic organizations, previously limited to on-premises infrastructure, can now consider AWS as a viable platform for their production workloads.

Getting started

We encourage customers to work with their Epic and AWS account teams to evaluate the r8i and r8ib instance families and determine the best cost-performance fit for your organization’s unique requirements.

Epic is a trademark of Epic Systems Corporation.

Kevin Vuong

Kevin Vuong

Kevin Vuong is a Global Healthcare Solution Architect at Amazon Web Services. With over 18 years of experience in healthcare, he is passionate about helping customers leverage AWS solutions to enhance the experiences of both clinicians and patients. In his free time, Kevin enjoys spending quality time with his wife and two children, engaging in activities like hiking, camping, and various sports.

Aaron Chen

Aaron Chen

Aaron Chen is a Principal Product Manager at Amazon EBS. At AWS, he is passionate about delivering high EBS performance on Amazon EC2 instances and building better data service experiences on Amazon EBS. Outside of work, he enjoys playing volleyball and reading books.

Abinash Nayak

Abinash Nayak

Abinash Nayak is a Senior Product Manager at Amazon Web Services (AWS). He has over 15 years of experience building and scaling technical solutions across diverse business domains, from finance and healthcare to startup ecosystems. At AWS, he manages high-memory Elastic Cloud Compute (EC2) products and specialized EC2 platforms. Outside of work, he enjoys spending time with family and staying active through sports.