What's New with AWS Lambda (CNS376)

Session Notes

Lambda's Key Differentiation: Speed

• Main aspiration: Provide speed to customers and builders
• Speed encompasses code generation and the entire shipping cycle
• Faster shipping leads to more engaging applications and user feedback
• Applications are well-architected by default, handling scalability, availability, and reliability
• Primary benefit: No infrastructure to manage, reducing operations and increasing speed

Launches Bucketed into 3 Cohorts

1 New Primitives

1. Introduction of Lambda's managed instances

2 Managed Instances

1. Mental model: Lambda shines with needlepoint workloads (high traffic spikes)
2. For steady-state traffic, builders seek to optimize performance and costs
3. Allows choice of compute instances, network, or memory for performance optimization
4. Retains full surveillance operations and developer experience
5. Event source integrations, patching, routing, and scaling are handled by Lambda
6. Introduces multi-concurrency for more efficient request handling
7. Incorporates EC2’s pricing incentives (savings plans, reserved instances) for cost

optimization

3 Capacity Provider

1. Additional requirement: Create a capacity provider

10.Specify preferences for compute instances, memory instances, or scaling profile

11.Optionally, let Lambda choose and continuously improve price performance

Use Cases and Features

• Steady State Applications: Seamlessly handled by Lambda
• Needlepoint Bursts/Traffic Management: Leave on Lambda for functionality
• Steady State/Popularity: Move to managed instances for optimization
• Performance Critical Apps: Specialized instances available
• Variety of Applications: Media data processing, web applications, event-driven applications
• Regulatory Requirements: Managed instances allow preferences for compute zones

Tenant Isolation Feature

• Use Case: scenario needing isolation between requests
• Benefit: Pass unique tenant ID or JWT token for clean, isolated execution environment
• Improves: CICD cycles and eliminates need for custom tooling

Well-Architected by Default

• Value Prop: Latest programming languages with performance and security fixes
• Recent Additions: Python 3.14, Java 2.5, Node.js 24
• Benefit: Improves developer productivity and enables safe, faster software shipping
• Mental Model: Make new runtimes available within 90 days of community release

Additional Features and Benefits

Runtime Upgrading

• Automated Patches: Handles vulnerabilities like Log4j
• Challenge: High effort to upgrade runtime for large function counts
• Solution: AWS Transform Custom (Gen AI-based upgrades)
• Benefit: Reduces tech debt by up to 85%, seamless integration into dev cycles
• Snapstart:
• Problem: Cold starts during function initialization
• Solution: Snapshots execution environment for faster subsequent invokes
• Benefit: Reduced cold start times, no code changes or custom tooling required
• Fault Injection Service (FIS) Integration:
• Purpose: Test application resilience under stress conditions
• Features: Specify conditions like increased latency or unavailable downstream services
• Benefit: Helps plan and prevent outages, increases confidence in production scenarios
• Improved Observability:
• Challenge: Lack of visibility into event source polling mechanisms
• Solution: Enabled additional CloudWatch metrics for count, log, and throttles
• Benefit: Instant issue detection and improved time to resolution, no custom tooling required

AWS Transform custom

AWS-managed, out-of-the-box transformations that are pre-built, AWS-vetted transformations

for common upgrade scenarios. These are ready to use without any additional setup.

Currently available transformations

• Java 8 to 17 migrations (for both Gradle and Maven)
• Node.js 12 to 22 upgrades (including Lambda environments)
• Python runtime updates to 3.11/3.12/3.13 (standard and Lambda)
• AWS SDK migrations (v1 to v2)
• Key characteristics:
• Validated by AWS - These transformations are vetted by AWS to be high quality
• Ready to use - No additional setup required
• Continuously growing - Additional transformations are continually being added
• Customizable - Pre-built transformations can be enhanced with specific rules for your organization's needs (e.g., adding rules for handling internal libraries or coding standards)
• Experimental support - Some transformations may be marked as experimental as they undergo further testing

Additional Features and Benefits

Additional CloudWatch Metrics

• Problem: Lack of visibility into event source polling mechanisms (Kafka, SQS)
• Solution: Enabled additional metrics for polar count, lag, and throttles
• Benefit: Instant issue detection and improved time to resolution, no custom tooling required
• Schema Registry Support for Avro Format:
• Problem: Builders had to manually add boilerplate code for Avro serialization/deserialization
• Solution: Added capability to auto serialize and deserialize Avro events
• Benefit: Less code, fewer errors, schema evolution support, no custom tooling required
• Provision Mode for SQS:
• Problem: Customers needed faster scaling and control over concurrency for polling
• Solution: Provision mode allows pre-warming capacity to handle spikes instantly
• Benefit: Eliminates delays during spikes, helps meet SLAs, optimizes costs, no custom tooling required
• Accelerating Developers:
• Goal: Help developers ship software faster
• Developer Preference: Develop in local machines/IDE after initial console testing
• Focus: Streamline development process to reduce time to market

Features to Accelerate Development Cycles

Seamless Console to IDE Transition

• Feature: Build bare-bones application on console with single click
• Benefit: Packages dependencies, lights up on local IDE for immediate coding without manual work
• Local Testing with LocalStack:
• Partner: LocalStack emulates AWS services (storage, database, networking)
• Benefit: Develop and test offline fully on local machine, leading to faster iterations of business logic without custom tooling
• Remote Debugging:
• Feature: Enable checkpoints/breakpoints with two clicks
• Benefit: Analyze variables and source code locally while code runs in production, test IAM policies, roles, database connectivity, VPC, and network configurations without custom tooling
• MCP Server for Best Practices:
• Feature: Bakes in best practices (input validations, error handling, status codes)
• Benefit: Generates better quality, consistent code, reduces code review cycles, accelerates velocity
• Support: Web apps and event-driven applications (Kafka triggers, event-driven architectures)