Amazon EC2 R9g and R9gd Instances Powered by AWS Graviton5 Now Generally Available
Amazon EC2 R9g and R9gd Instances Powered by AWS Graviton5 Processors Generally Available
Designed specifically for memory-intensive enterprise workloads, the new R9g silicon delivers up to 25% higher compute performance per vCPU compared to its predecessor, the Graviton4-based R8g, while operating as the most energy-efficient processor built by AWS to date.
The Tech TL;DR:
- Silicon Upgrade: Powered by AWS Graviton5 processors, featuring DDR5 8800 MT/s memory and a 5x larger L3 cache for significantly improved data locality.
- Workload Targets: Tailored for memory-bound applications like Redis, Valkey, MemCached, relational databases, real-time big data analytics, and containerized microservices running on Kubernetes or Docker.
- Deployment Scale: Available now in 11 sizes ranging from medium to metal-48xl, featuring Instance Bandwidth Configuration (IBC) and the Nitro Isolation Engine for mathematically proven hardware security.
Architectural Breakdown and Hardware Specs of AWS Graviton5
This bandwidth expansion is paired with an L3 cache sized up to 5 times larger than previous generations, drastically cutting down cache-miss penalties for high-throughput databases and in-memory key-value stores.
Network and storage subsystems receive an equally steep modernization. The largest instance variants scale up to 100 Gbps of network bandwidth and 72 Gbps of Amazon EBS bandwidth, supported by up to 3x higher packet-processing performance.
Furthermore, R9g and R9gd instances support Instance Bandwidth Configuration (IBC). This capability allows systems engineers to dynamically adjust the bandwidth allocation between Amazon EBS and Amazon VPC networking by 25%. Such granular control helps optimize mixed workloads that place concurrent stress on block storage and network interfaces.
Comparative Instance Specifications
The R9g family is available in 11 distinct sizing tiers, stretching from single-vCPU entry points up to bare-metal configurations. R9gd variants supply identical compute and networking throughput while integrating local NVMe-based SSD block-level storage for low-latency scratch space, local caches, and open-source database nodes.
| Instance Size | vCPUs | Memory (GiB) | Instance Storage (R9gd) | Network (Gbps) | EBS (Gbps) |
|---|---|---|---|---|---|
| r9g.medium / r9gd.medium | 1 | 8 | EBS-Only / 1 x 59 GB | Up to 15 | Up to 12 |
| r9g.xlarge / r9gd.xlarge | 4 | 32 | EBS-Only / 1 x 237 GB | Up to 15 | Up to 12 |
| r9g.8xlarge / r9gd.8xlarge | 32 | 256 | EBS-Only / 1 x 1900 GB | 17 | 12 |
| r9g.24xlarge / r9gd.24xlarge | 96 | 768 | EBS-Only / 3 x 1900 GB | 50 | 36 |
| r9g.48xlarge / r9gd.48xlarge | 192 | 1536 | EBS-Only / 3 x 3800 GB | 100 | 72 |
Hypervisor Security and the Nitro Isolation Engine
All R9g and R9gd instances execute on the AWS Nitro System, which offloads virtualization, storage, and networking tasks to dedicated custom hardware. This design grants virtualized environments near-bare-metal performance. Security isolation is enforced via the Nitro Isolation Engine, a component introduced to the C9g and M9g line earlier in the year.
Per published AWS technical whitepapers, the Nitro Isolation Engine uses formal verification—a mathematical technique used to prove that hardware and software logic behave strictly as intended across all possible states, rather than merely passing empirical test suites. The engine mediates all access to virtual machine memory, CPU register states, and I/O hardware through a minimal API surface. This mechanism establishes the Nitro hypervisor as a formally verified execution layer, addressing critical compliance frameworks like SOC 2 and ISO 27001.
Migration Paths and Developer Tooling
For engineering organizations migrating workloads from R8g instances, AWS reports that no source code alterations are required for the vast majority of applications. Multi-arch container images compiled for Arm64 run natively across Amazon EKS, Amazon ECS, and self-managed Kubernetes environments. Standard Linux distributions, including Ubuntu 22.04+, RHEL 8.4+, Debian 12+, and Amazon Linux 2023, feature out-of-the-box support for the new silicon.
Teams modernizing legacy codebases can leverage automated migration utilities. Developers working in enterprise Java environments can utilize the AWS Transform tool to automate code refactoring from x86 to Graviton targets. Integration scripts can be validated via the command line using standard container build pipelines:
docker buildx build --platform linux/arm64 -t my-app:graviton5 --push .Regional Availability and Deployment Options
Amazon EC2 R9g and R9gd instances are immediately accessible in US East (N. Virginia, Ohio), US West (Oregon), and Europe (Frankfurt) AWS Regions. Compute capacity can be procured via On-Demand instances, Savings Plans, Spot Instances, Dedicated Instances, or Dedicated Hosts.
Frequently Asked Questions
What programming languages are supported on AWS Graviton5 instances?
Graviton5 instances support a broad spectrum of languages and runtimes, including C/C++, Rust, Go, Java, Python, .NET Core, Node.js, Ruby, and PHP, provided they are compiled or run within an Arm64-compatible environment.
How does the Nitro Isolation Engine secure multi-tenant execution?
The Nitro Isolation Engine uses formal verification to mathematically prove hardware and software isolation boundaries between virtual machines, strictly mediating access to memory, CPU registers, and I/O devices.