- Newest
- Most votes
- Most comments
ENA is enabled by default with supported instances on certain Windows/Linux kernels launched with an AMI from Amazon. Referenced here: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking-ena.html - Linux https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/enhanced-networking-ena.html - Windows
If you run the command aws ec2 describe-instances --instance-ids instance_id --query "Reservations[].Instances[].EnaSupport" and it shows true, and the driver is installed for "Elastic Network Adapter", you have ENA enabled.
As for the 100Gb question, you'll need an instance that supports 100Gb/s. There aren't many, but they are in tables for each type of instance below. Accelerated Computing: https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/accelerated-computing-instances.html#gpu-network-performance Compute Optimized: https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/compute-optimized-instances.html#compute-network-performance General Purpose: https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/general-purpose-instances.html#general-purpose-network-performance Memory Optimized: https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/memory-optimized-instances.html#memory-network-perf Storage Optimized: https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/storage-optimized-instances.html#storage-network-performance
100Gb/s bandwidth means aggregate bandwidth between instances. Single flow bandwidth can only go beyond 5Gb/s up to 25Gbps for instances that support ENA Express. https://docs.aws.amazon.com/AWSEC2/latest/WindowsGuide/ena-express.html
Relevant content
- Accepted Answerasked 2 years ago
- asked 25 days ago
AWS OFFICIALUpdated 8 months ago- AWS OFFICIALUpdated 4 months ago
- AWS OFFICIALUpdated 2 years ago
In relation to this last part, "...Single flow bandwidth can only go beyond 5Gb/s up to 25Gbps for instances that support ENA Express...."
I don't understand if this is a general constraint for all types of instances and if it is independent of the specific bandwidth indicated for the instance. Additionally, I'm trying to ask this: I am conducting tests between instances in different regions and also between 2 instances in us-east-1 and us-west-2, but I can't figure out if I can specify in my analyses the bandwidth as indicated in the links above or if there are different network performance based on whether we are in the same region or not. For example, in this link (https://aws.amazon.com/blogs/networking-and-content-delivery/improving-performance-on-aws-and-hybrid-networks/), I read:
"Traffic to other AWS Regions, an Internet Gateway (IGW), Virtual Private Gateway (VPG), or Local Gateway (used for AWS Outposts) can utilize up to 50% of the network bandwidth available to a current generation instance with a minimum of 32 vCPUs, or 5 Gbps for a current generation instance with less than 32 vCPUs."
On this last part, when it says 5Gbps with less than 32 vCPUs, is this a statement that applies to all types of instances, even if they are indicated to have performance higher than 5Gbps?
Please help me understand these network aspects, as otherwise, my measurements with iperf3 don't make much sense. The goal is that for the measurements I make between different regions, I ne