CloudPro #14: Agents for Amazon Bedrock, Google Document AI Workbench, Dockerfile deployments on Koyeb, Blue Green deployment in Azure Cloud
Bite-sized actionable content, practical tutorials, and resources for cloud professionals.
Welcome to a brand new edition of the CloudPro! In this issue, we're turning our focus to a new product launched by AWS called "Agents for Amazon Bedrock". Join us as we take a closer look at this product and understand its capabilities.
In this issue, we also highlight Google’s Document AI Workbench, Dockerfile deployments on Koyeb, Snowflake’s External Access in Snowpark, and new Amazon EC2 R7iz instances.
We know how much you love your weekly dose of our handpicked cloud resources, tutorials, and secret knowledge. That’s why we also discuss how to use Terraform, setup monitoring in Kubernetes, measure Github Copilot’s impact on engineering productivity, and Blue-Green Deployment in Azure Cloud.
But that’s not all! We’ve also curated the best cloud GitHub repos for you to use and test like OpenFunction, the Cloud Native Function-as-a-Service Platform, and a batteries-included Python client library for Kubernetes called kr8s.
Cheers,
Shreyans Singh
Editor-in-Chief
PS: I hope you will enjoy today's newsletter! I’m all ears for your thoughts – the good, the great, and the "meh." Share your feedback and snag a free Packt eBook (Mastering Kubernetes by Gigi Safyan). It's a win-win. Can't wait to hear what you think! Share your feedback and get a free Packt eBook!
⚡ TechWave: Cloud News & Analysis
⚡Connect Foundation Models to Your Company Data Sources with Agents for Amazon Bedrock: AWS has launched a new product called "agents for Amazon Bedrock" that helps developers create AI applications. These applications can understand and respond to user queries. The feature allows these applications to access additional information from a company's data without needing to retrain the AI model. This helps in giving more accurate and relevant responses.
To enable this, developers can create a "knowledge base" by storing their data in a specific way. This process involves converting the data into special numerical representations (called embeddings) and storing them in a special database. This allows the AI application to quickly retrieve relevant information when a user asks a question.
This method, called "Retrieval Augmented Generation" (RAG), is useful because it allows the AI to access up-to-date information without constantly retraining the model. It's particularly beneficial for tasks like answering questions about specific topics.
⚡Document AI Workbench: Custom Extractor and Summarizer: Google Cloud has introduced new generative AI-powered features to Document AI Workbench, a tool that helps businesses process and analyze documents more efficiently. These features include:
Custom Extractor with Generative AI: This tool allows you to extract data from documents, even when they contain free-form text, complex layouts, or have minimal training data available.
It offers several advantages: faster results, customize the results, and handle larger documents.
To use it, you create a Custom Extractor, define fields you want to extract, and upload a sample document for preview. You can confirm or correct predictions to improve accuracy, and the model learns from your input.
The Summarizer: Provides document summaries and doesn't require any training. You can customize summaries based on length and format preferences:
Easy setup: You can create a Summarizer, call the endpoint with a document, and get summaries.
Customization: You can dictate length and format settings when calling the API per document.
⚡Dockerfile Deployment on High-Performance MicroVMs is GA: Koyeb now supports Dockerfile-based deployments, allowing you to deploy various types of applications easily.
Dockerfile-based deployments offer more flexibility. You can use Dockerfiles to deploy various types of applications, frameworks, and runtimes, including .NET, Rust, Zig, Bun, Crystal, Deno, and more.
Koyeb offers three different methods for deploying applications close to your users:
Build from Dockerfile: Deploy GitHub repositories containing a Dockerfile and benefit from continuous deployment with a simple git push.
Cloud Native Buildpacks: Deploy GitHub repositories with code in various languages, and Koyeb will automatically detect your runtime and build it.
Pre-built Containers: Deploy pre-built container images from any registry with a single API call and maintain control over the build process.
You can customize Docker builds using environment variables and build arguments for specific configurations. Koyeb automatically caches each layer when building Dockerfiles. This means that when you push a new version of your application, Koyeb can reuse the cached layers to speed up the build process.
🌐From the Cloud World:
🌐Google Cloud's Bigtable new change stream feature let you track real-time data changes in Cloud Bigtable, making it useful for various purposes like analytics, event-based apps, migrations, and compliance.
🌐New Amazon EC2 R7iz instances are powerful cloud-based virtual machines optimized for high CPU performance and memory-intensive tasks, making them ideal for applications like electronic design, databases, and data analytics, while offering various sizes and accelerators for different workload needs.
🌐External Access in Snowpark allows users to securely connect to external network locations, such as APIs or services on the internet, directly from their Snowpark code within Snowflake, enhancing flexibility and functionality while maintaining high security and governance.
🌐The VMware Partner Program for Object Storage Extension allows software vendors to integrate their solutions with VMware's Cloud Director Object Storage Extension (OSE), enhancing cloud storage capabilities and user experience.
🌐JFrog has extended its CI/CD platform to include security features and ML model management, enabling DevOps teams to securely integrate and manage AI models within their workflows.
⭐ MasterClass: Tutorials & Guides
⭐Self-hosted GitHub runners on AWS: This article explains how to set up self-hosted GitHub runners on AWS, which are used for automating tasks in software development, to save costs, access private networks, and customize the environment, and it discusses three approaches: hosting runners on EC2 instances, scaling with auto-scaling, and an event-driven solution for secure and scalable infrastructure.
⭐Blue-Green Deployment in Azure Cloud. A Step-by-step Guide: This is a method for updating software applications without causing downtime or risking bugs by running two identical versions (Blue for current and Green for new) simultaneously and gradually transitioning users to the new version once it's confirmed to work correctly.
⭐Stable diffusion containers on gke: The article discusses how to improve the startup time of the Stable Diffusion model on Google Kubernetes Engine (GKE) by up to 4x by optimizing the Dockerfile, separating base environment from runtime dependencies, leveraging high-performance disk images, and using GKE Image Streaming and Cluster Autoscaler.
⭐Kubernetes Part 20: Deploy Prometheus and Grafana: This blog explains how to set up monitoring in Kubernetes using Prometheus (data collector) and Grafana (dashboard) by creating persistent volumes, configuring modules, and deploying the monitoring components.
⭐Snowpark ML: The ‘Easy Button’ for Open Source LLM Deployment in Snowflake: This blog discusses how Snowflake enables the deployment of large language models (LLMs), specifically Meta's Llama2, within Snowflake using Snowpark ML, the Snowpark Model Registry, and Snowpark Container Services, allowing users to apply generative AI transformations to their data seamlessly.
🔍Secret Knowledge: Learning Resources
🔍Measuring GitHub Copilot's Impact on Engineering Productivity: This article discusses how to measure the impact of GitHub Copilot on engineering productivity, and it provides key metrics to quantify this impact, such as velocity, time to value, quality, dependability, and developer satisfaction, and explains how to combine these metrics to create a business case for GitHub Copilot based on productivity improvements.
🔍Kubernetes Cluster Management and Cloud Automation: Projectsveltos, Crossplane, and ClusterAPI are three open-source projects that work together to simplify the management of Kubernetes clusters and cloud resources, allowing you to automate and streamline the process of creating and managing infrastructure in Kubernetes.
🔍What is a cloud security assessment? This article provides a 10-point checklist for conducting a comprehensive cloud security assessment, including steps like understanding the current state of your cloud environment, identifying assets, classifying data, assessing threats and vulnerabilities, and implementing controls, emphasizing the importance of maintaining ongoing vigilance in cloud security.
🔍AWS Gurus, Battle Tested Processes, On Your Team: This article discusses Amazon EventBridge Pipes, a feature introduced in December 2022, explaining how they can be used to connect AWS resources and highlighting scenarios where they are relevant and how they differ from traditional approaches.
🔍Optimize AWS Storage Costs with S3 Lifecycle Configurations: This article explains how to optimize storage costs on Amazon Web Services (AWS) using Amazon S3 Lifecycle Configurations, which are rules that automate actions like moving old data to lower-cost storage tiers or deleting it, ultimately saving money on AWS bills.
🛠️HackHub: Trending Tools
🛠️taubyte/dreamland: run a Serverless Cloud on your laptop
🛠️kr8s-org/kr8s: A batteries-included Python client library for Kubernetes
🛠️powerfulseal/powerfulseal: A powerful testing tool for Kubernetes clusters.
🛠️kumahq/kuma: The multi-zone service mesh for containers, Kubernetes and VMs
🛠️OpenFunction/OpenFunction: Cloud Native Function-as-a-Service Platform
The Terraform Cookbook is a guide that teaches you how to use Terraform, a tool for managing data center infrastructure through code, with examples for Azure, AWS, and GCP, covering testing, automation, Docker, Kubernetes, GitOps, and troubleshooting.
Read the first chapter for free!
📢 If your company is interested in reaching an audience of developers and, technical professionals, and decision makers, you may want to advertise with us.
If you have any comments or feedback, just reply back to this email.
Thanks for reading and have a great day!