I always seek good resources for my teaching at #universityofoxford and this is a great free book on Cloud native – a topic that is very much in the focus
The book needs a free registration and is provided as a full ebook by d2iq.
D2iq was formerly mesosphere – the commercial arm of mesos – but now is focussed towards Kubernetes and cloud native services. See this techcrunch link for some background.
Previously, some chapters were free. Now the whole book seems to be a free download
Containers and Cloud native is the way of the future. By 2023, 70% of applications deployed in the cloud will use containers as a packaging mechanism, up from 20% in 2020. (vmware)
The book table of contents is
Introduction to Cloud Native
- Distributed Systems
- Fallacies of Distributed Systems
- CAP Theorem
- The Twelve-Factor App
- Availability and Service-Level Agreements
- Summary
Fundamentals
- Containers
- Container Isolation Levels
- Container Orchestration
- Kubernetes Overview
- Kubernetes and Containers
- Serverless Computing
- Functions
- From VMs to Cloud Native
- Lift-and-Shift
- Application Modernization
- Application Optimization
- Microservices
- Benefits of a Microservices Architecture
- Challenges with a Microservices Architecture
- Summary
Designing Cloud Native Applications
- Fundamentals of Cloud Native Applications
- Operational Excellence
- Security
- Reliability and Availability
- Scalability and Cost
- Cloud Native versus Traditional Architectures
- Functions versus Services
- Function Scenarios
- Considerations for Using Functions
- Composite of Functions and Services
- API Design and Versioning
- API Backward and Forward Compatibility
- Semantic Versioning
- Service Communication
- Protocols
- Messaging Protocols
- Serialization Considerations
- Idempotency
- Request/Response
- Publisher/Subscriber
- Choosing Between Pub/Sub and Request Response
- Synchronous versus Asynchronous
- Gateways
- Routing
- Aggregation
- Offloading
- Implementing Gateways
- Egress
- Service Mesh
- Example Architecture
- Summary
Working with Data
- Data Storage Systems
- Objects, Files, and Disks
- Databases
- Streams and Queues
- Blockchain
- Selecting a Datastore
- Data in Multiple Datastores
- Change Data Capture
- Write Changes as an Event to a Change Log
- Transaction Supervisor
- Compensating Transactions
- Extract, Transform, and Load
- Microservices and Data Lakes
- Client Access to Data
- Restricted Client Tokens (Valet-Key)
- Database Services with Fine-Grained Access Control
- GraphQL Data Service
- Fast Scalable Data
- Sharding Data
- Caching Data
- Content Delivery Networks
- Analyzing Data
- Streams
- Batch
- Data Lakes on Object Storage
- Data Lakes and Data Warehouses
- Distributed Query Engines
- Databases on Kubernetes
- Storage Volumes
- StatefulSets
- DaemonSets
- Summary
DevOps
- What Is DevOps?
- Collaboration
- Automation
- Lean Principles and Processes
- Measurement
- Sharing
- Testing
- Test Doubles
- Test Automation Pyramid
- When to Run Which Types of Tests
- Testing Cadence
- Testing in Production
- Development Environments and Tools
- Development Tools
- Development Environments
- Local Development Environments
- Table of Contents | v
- Local Development with a Remote Cluster
- Skaffold Development Workflow
- Remote Cluster Routed to Local Development
- Cloud Development Environments
- CI/CD
- Source Code Control
- Build Stage (CI)
- Test Stage (CI)
- Deploy Stage (CD)
- Release Stage (CD)
- Post-Release Stage
- Monitoring
- Collecting Metrics
- Observable Services
- Configuration Management
- Single-Environment Variable
- Multiple-Environment Variables
- Adding ConfigMap Data to a Volume
- Storing Secrets
- Deployment Configuration
- Sample CI/CD Flows
- Summary
Best Practices
- Moving to Cloud Native
- Breaking Up the Monolith for the Right Reasons
- Decouple Simple Services First
- Learn to Operate on a Small Scale
- Use an Anticorruption Layer Pattern
- Use a Strangler Pattern
- Come Up with a Data Migration Strategy
- Rewrite Any Boilerplate Code
- Reconsider Frameworks, Languages, Data Structures, and Datastores
- Retire Code
- Ensuring Resiliency
- Handle Transient Failures with Retries
- Use a Finite Number of Retries
- Use Circuit Breakers for Nontransient Failures
- Graceful Degradation
- Use a Bulkhead Pattern
- Implement Health Checks and Readiness Checks
- Define CPU and Memory Limits for Your Containers
- Implement Rate Limiting and Throttling
- Ensuring Security
- Treat Security Requirements the Same as Any Other Requirements
- Incorporate Security in Your Designs
- Grant Least-Privileged Access
- Use Separate Accounts/Subscriptions/Tenants
- Securely Store All Secrets
- Obfuscate Data
- Encrypt Data in Transit
- Use Federated Identity Management
- Use Role-Based Access Control
- Isolate Kubernetes Pods
- Working with Data
- Use Managed Databases and Analytics Services
- Use a Datastore That Best Fits Data Requirements
- Keep Data in Multiple Regions or Zones
- Use Data Partitioning and Replication for Scale
- Avoid Overfetching and Chatty I/O
- Don’t Put Business Logic in the Database
- Test with Production-like Data
- Handle Transient Failures
- Performance and Scalability
- Design Stateless Services That Scale Out
- Use Platform Autoscaling Features
- Use Caching
- Use Partitioning to Scale Beyond Service Limits
- Functions
- Write Single-Purpose Functions
- Don’t Chain Functions
- Keep Functions Light and Simple
- Make Functions Stateless
- Separate Function Entry Point from the Function Logic
- Avoid Long-Running Functions
- Use Queues for Cross-Function Communication
- Operations
- Deployments and Releases Are Separate Activities
- Keep Deployments Small
- CI/CD Definition Lives with the Component
- Consistent Application Deployment
- Use Zero-Downtime Releases
- Don’t Modify Deployed Infrastructure
- Use Containerized Build
- Describe Infrastructure Using Code
- Use Namespaces to Organize Services in Kubernetes
- Isolate the Environments
- Separate Function Source Code
- Correlate Deployments with Commits
- Logging, Monitoring, and Alerting
- Use a Unified Logging System
- Use Correlation IDs
- Include Context with Log Entries
- Common and Structured Logging Format
- Tag Your Metrics Appropriately
- Avoid Alert Fatigue
- Define and Alert on Key Performance Indicators
- Continuous Testing in Production
- Start with Basic Metrics
- Service Communication
- Design for Backward and Forward Compatibility
- Define Service Contracts That Do Not Leak Internal Details
- Prefer Asynchronous Communication
- Use Efficient Serialization Techniques
- Use Queues or Streams to Handle Heavy Loads and Traffic Spikes
- Batch Requests for Efficiency
- Split Up Large Messages
- Containers
- Store Images in a Trusted Registry
- Utilize the Docker Build Cache
- Don’t Run Containers in Privileged Mode
- Use Explicit Container Image Tags
- Keep Container Images Small
- Run One Application per Container
- Use Verified Images from Trusted Repositories
- Use Vulnerability Scanning Tools on Images
- Don’t Store Data in Containers
- Never Store Secrets or Configuration Inside an Image
- Summary
Portability
- Why Make Applications Portable?
- The Costs of Portability
- Data Gravity and Portability
- When and How to Implement Portability
- Standardized Interfaces
You can download the whole book for a free registration HERE
Credit: Data Science Central By: ajit jaokar
