Highlights from this week (mostly cloud with a bit of Docker and CSS thrown in):
AWS Bedrock
—AWS is building new tools that will encourage and ease the adoption of generative AI. These tools will provide cost-effective infrastructure for model training and emphasise building secure application. Specifically, Amazon Bedrock is now generally available. Bedrock is a managed service for building generative AI applications. It simplifies the development process while maintaining privacy and security. You can quickly get started with Bedrock by experimenting directly in the AWS Management Console. Alternatively, it’s accessible via LangChain or from Python using the boto3 package.

EC2 C7a Instances
—AWS has launched C7a instances on EC2. These instances can achieve a peak frequency of 3.7 GHz and are up to a 50% more performant than the C6a instances launched in February 2022. Memory bandwidth is enhanced by DDR5. The C7a instances are a performance-optimised, cost-effective and energy-efficient option for compute-intensive tasks.
| Name | vCPUs | Memory (GiB) | Network Bandwidth (Gbps) | EBS Bandwidth (Gbps) |
|---|---|---|---|---|
| c7a.medium | 1 | 2 | Up to 12.5 | Up to 10 |
| c7a.large | 2 | 4 | Up to 12.5 | Up to 10 |
| c7a.xlarge | 4 | 8 | Up to 12.5 | Up to 10 |
| c7a.2xlarge | 8 | 16 | Up to 12.5 | Up to 10 |
| c7a.4xlarge | 16 | 32 | Up to 12.5 | Up to 10 |
| c7a.8xlarge | 32 | 64 | 12.5 | 10 |
| c7a.12xlarge | 48 | 96 | 18.75 | 15 |
| c7a.16xlarge | 64 | 128 | 25 | 20 |
| c7a.24xlarge | 96 | 192 | 37.5 | 30 |
| c7a.32xlarge | 128 | 256 | 50 | 40 |
| c7a.48xlarge | 192 | 384 | 50 | 40 |
| c7a.metal-48xl | 192 | 384 | 50 | 40 |
IP4 Charges
—From February 2024 AWS will charge for each public IPv4 addresses at a rate of $0.005 per IP per hour (that’s around $3.60 per month). Why? Because IP addresses are becoming a scarce resource. See this for suggestions on how to monitor and optimise these costs.
Docker AI
—Docker AI was announced at DockerCon earlier this month. Trained on collective best practices, Docker AI will offer guidance on editing or debugging Docker files. It should promote up-to-date and secure application images and allow developers to focus attention on their application rather than on tools and infrastructure.