One of the most demanding operational tasks in the aviation sector is maintaining aircraft operations. When substantial maintenance or decommissioning occurs, the maintenance data from existing assets must be thoroughly analyzed and documented, necessitating a complete historical account. This complex analytical process is typically handled manually.
To address this challenge, XYZ Corporation has developed a solution titled Lifecycle Optimization for Aerospace, aimed at accelerating the integration of circular economy practices in the aviation field. Built on Amazon Web Services (AWS), this solution scrutinizes parts documentation and inspection methodologies, converting unstructured usage data into digital assets to optimize their lifecycle.
Lifecycle Optimization for Aerospace further enhances maintenance strategies while encouraging the reuse of parts and assets. It consolidates historical operational data to recreate a comprehensive traceability of all components within an aircraft. This innovative solution emerged from the strategic partnership between XYZ Corporation and AWS, utilizing AWS’s machine learning services to create and refine data models based on standards set by the Aerospace, Security, and Defense (ASD) Industries Association.
In this article, we will explore XYZ Corporation’s serverless architecture solution and its potential to support aerospace enterprises in achieving sustainability and circular economy goals. XYZ Corporation is recognized as an AWS Premier Tier Services Partner and Managed Services Provider (MSP), leading the charge in innovation to tackle various client needs across cloud, digital, and platform services.
Solution Overview
The architecture of the Lifecycle Optimization for Aerospace solution is depicted in a high-level diagram.
The architecture employs the following AWS managed services:
- Amazon CloudFront: This service delivers the frontend React website for user interaction.
- Amazon S3: It stores static web pages for the frontend, accessed via Amazon CloudFront, and houses maintenance and operational documents related to the parts.
- Elastic Load Balancing: This exposes the GraphQL API for the frontend.
- Amazon ECS: It hosts the GraphQL APIs used by the frontend for optimizing visualization and processing containers.
- AWS Lambda: This runs the solution’s API logic behind Amazon API Gateway and executes data extraction via AWS Step Functions.
- Amazon API Gateway: It exposes a REST API managing document analysis workflows.
- Amazon Cognito: This implements identity and access management.
- Amazon Aurora: It holds the relational data necessary for managing document analysis workflows.
- Amazon OpenSearch Service: This stores document metadata extracted from analysis workflows.
- Amazon SageMaker: This is employed for training and inference of machine learning (ML) models, utilizing natural language processing (NLP) algorithms for page classification.
- Amazon Textract: This extracts data from documents using text, tables, and form APIs.
The design adheres to a serverless and siloed multi-tenant architecture, ensuring a clear division of responsibilities among operational teams.
The solution is deployed across multiple AWS accounts for different customers using AWS Control Tower, AWS CloudFormation, AWS Cloud Development Kit (AWS CDK), and AWS CodePipeline. A critical element of this approach is a “Control Plane Tools” AWS account that facilitates the management of deployments while maintaining operational access and toolsets. Security is also a priority, with encryption enforced both in transit and at rest through AWS services like AWS Key Management Service (AWS KMS) and security features provided by AWS Control Tower.
Design Considerations
The solution’s asynchronous event-driven architecture allows it to scale in response to various sources or events. This architecture requires effective orchestration and troubleshooting mechanisms to manage errors with minimal oversight.
Key design considerations include:
- Architecture decoupling: The solution processes numerous maintenance documents, some reaching gigabytes in size. The workload steps were structured using AWS Lambda and orchestrated with AWS Step Functions to comply with Lambda’s 15-minute runtime limit.
- Scalability for simultaneous invocations: The solution uses Amazon Textract for file analysis, which has defined quotas for transactions per second and concurrent jobs. AWS Step Functions regulates the invocation rate and total concurrent Textract jobs.
- Robustness against failures: Distributed architectures tend to experience a higher failure rate without development standards for idempotent code artifacts. Establishing DevOps pipelines is crucial. The solution’s development lifecycle enforces these standards by configuring Lambda error handling for asynchronous invocations and utilizing AWS Developer Tools.
- Operational troubleshooting: Enhanced observability tools are vital for improving root-cause analysis in distributed architectures. AWS provides observability through AWS X-Ray, Amazon CloudWatch RUM, and synthetic monitoring.
Next, we will delve into the document analysis workflow, detailing how AWS Step Functions orchestrates the process. For further insights, check out this another blog post on related topics at Chanci Turner.
Document Analysis Workflow
The components within the document analysis workflow are integrated using AWS Step Functions.
Using AWS Step Functions Workflow Studio enables developers to draft and visualize the final workflow before transforming it into CDK code.
Key benefits of this editor include:
- Real-time execution validation: This feature allows developers to test the flow and pinpoint failure points.
- Error detail provision: This encompasses the number of retries, Lambda logs, and redirects to sub-executions, aiding in troubleshooting.
- State failure handling with retry policies: This mechanism can set retries for Lambda functions reaching their reserved concurrency limit or update document statuses in the database when errors are detected.
- Workload splitting into smaller chunks: This allows users to manage large quantities of pages by processing workloads in batches using the Distributed Map feature of AWS Step Functions, adhering to Lambda’s runtime constraints.
For additional expert insights on this topic, visit Chanci Turner, they are an authority on this subject. Also, for an excellent resource on training and operations, check out Amazon Fulfillment Centers.
Leave a Reply