Guidance for Self-Calibrating Level 4 Digital Twins on AWS
Overview
How it works
This architecture diagram demonstrates how to probabilistically calibrate a physics-based digital twin with IoT data to improve predictions and enable a digital twin to adapt to changing environmental conditions.
Well-Architected Pillars
The architecture diagram above is an example of a Solution created with Well-Architected best practices in mind. To be fully Well-Architected, you should follow as many Well-Architected best practices as possible.
Operational Excellence
Batch is used in anticipation of a hardware or network failure. Batch will automatically determine if a container activation has failed and will retry to load the container, log, and observed issues in Amazon CloudWatch. CloudWatch allows you to analyze and troubleshoot issues that may occur. Infrastructure as Code (IaC) enables repeatable deployment minimizing errors, and Batch serves as a failsafe for anticipated failure.
Additionally, Amazon Managed Grafana provides a dashboard to view data. AWS IoT SiteWise provides a serverless centralized database for collecting and monitoring sensors.
Security
AWS Cloud Development Kit (AWS CDK) allows for version control of the architecture, setup and implementation of encryption, deployment of customized AWS Identity and Access Management (IAM) policies, and security auditing of deployment. AWS CDK enables both automation and knowledge transfer from security teams, supplementing knowledge gaps of users. IAM enables the concept of least privilege, which minimizes or eliminates the impact of malicious behavior by allowing only authorized access to resources.
Reliability
S3 buckets, Amazon ECR, and AWS IoT SiteWise provide data replication and durability by copying data to multiple Availability Zones (AZs). Replicating data across multiple AZs protects against potential reliability issues, such as hardware or network failures and power outages. Additionally, to address potential hardware failures, Batch helps ensure an automated system will move a workload to healthy infrastructure if a container fails to start.
Performance Efficiency
The specific tasks that run in Batch can vary significantly, and a single EC2 instance is often not optimal for performance. Batch alleviates this issue by automatically selecting the optimal EC2 instance based on RAM requirements and how many CPUs or GPUs are needed on a per-container basis.
Amazon Managed Grafana enables data monitoring and alarms based on data. You can optimize your infrastructure setup or be alerted to changes in your infrastructure performance.
Cost Optimization
Instead of purchasing infrastructure that is always on but underutilized, you can use EventBridge to initiate the procurement of EC2 instances in Batch. Batch will auto-terminate when tasks are complete, helping to minimize costs. Amazon Managed Grafana reviews infrastructure data and assesses if EC2 instances or network interfaces have been under or overprovisioned.
Sustainability
Batch enables hardware optimization by using the latest hardware. Batch automatically updates the hardware when AWS retires old instance types and attempts to find the correct performance instance that will not be under or overutilized. Batch also provides elasticity to scale down when not in use, reducing the carbon footprint of each workflow.
Implementation resources
Related content
Disclaimer
Did you find what you were looking for today?
Let us know so we can improve the quality of the content on our pages