How to replicate Amazon Web Services IoT SiteWise resources across environments

Introduction

As you scale your Amazon Web Services IoT SiteWise applications and move them into production, you may consider adopting common CI/CD methodologies that separate development and QA environments from production environments. This separation allows you to automate the deployment of these applications through deployment pipelines. You also may have multiple business units and/or industrial sites with common asset models and hierarchies that you would like to share and reuse across your organization. In these cases, customers typically have different Amazon Web Services accounts to separate environments, whether between dev and prod or between different business units. The following diagram depicts such an example where development is separated from QA and production environments:

To help customers replicate Amazon Web Services IoT SiteWise resources between environments, we created Amazon Web Services IoT SiteWise Tools , a set of utilities that allow you export Amazon Web Services IoT SiteWise Asset Models, Assets , and Amazon Web Services IoT SiteWise Monitor Dashboards into Amazon Web Services CloudFormation templates . The exported templates can then be used to recreate the exported resources into another Amazon Web Services environment. In this blog, you will see a sample walkthrough of how to use Amazon Web Services IoT SiteWise Tools to export models followed by an example architecture of how to automate the export and replication process in a CI/CD pipeline.

Asset Model Export Walkthrough

The utilities in the Amazon Web Services IoT SiteWise Tools repository give you the flexibility to replicate only the resources you need for your specific use case. You can choose to only export Amazon Web Services IoT SiteWise asset models, or also export the corresponding assets and Amazon Web Services IoT SiteWise Monitor dashboards. The export tool can be used manually from the command line (e.g. for a one-time export of an asset model into another environment) or can be integrated into your automation pipelines for CI/CD deployment scenarios. The utility can also be used to copy Amazon Web Services IoT SiteWise resources for multi-region deployments within the same account. The Amazon Web Services IoT SiteWise Tools repository has detailed documentation on how to use each of the utilities but for a basic demonstration of the tools, we created two asset models of a CNC Machine and Production Line as seen below. Each model contains a property and a hierarchical relationship between the two models.

To keep it simple, we will only export the models. Using the Amazon Web Services IoT SiteWise export tools, we optionally specify the region we want to export models from and run the command with no other flags (if you also want to export assets along with the models, you would simply add the -a, --assets flag). The command output will look something like the following:

If the command succeeds, a CloudFormation template will be saved to a folder in the local directory named cfnexport. In our example case the CloudFormation will look like the following:

{
    "AWSTemplateFormatVersion": "2010-09-09",
    "Description": "SiteWise Export",
    "Resources": {
        "CNCMachineResource": {
            "Type": "AWS::IoTSiteWise::AssetModel",
            "Properties": {
                "AssetModelName": "CNC Machine",
                "AssetModelProperties": [
                    {
                        "Name": "SpindleSpeed",
                        "DataType": "DOUBLE",
                        "Unit": "RPM",
                        "Type": {
                            "TypeName": "Measurement"
                        },
                        "LogicalId": "SpindleSpeed9f2e03dd"
                    }
                ],
                "AssetModelHierarchies": []
            }
        },
        "ProductionLineResource": {
            "Type": "AWS::IoTSiteWise::AssetModel",
            "Properties": {
                "AssetModelName": "Production Line",
                "AssetModelProperties": [
                    {
                        "Name": "Location",
                        "DataType": "STRING",
                        "Type": {
                            "TypeName": "Attribute",
                            "Attribute": {}
                        },
                        "LogicalId": "Locationafc85231"
                    }
                ],
                "AssetModelHierarchies": [
                    {
                        "Name": "CNC Machines",
                        "ChildAssetModelId": {
                            "Ref": "CNCMachineResource"
                        },
                        "LogicalId": "CNCMachines"
                    }
                ]
            }
        }
    }
}

This CloudFormation template can now be launched in another region or another Amazon Web Services Account to create the same asset models we defined above.

That’s it, now you have an understanding how the export utility works. In the next section we will provide an example architecture that shows how you can integrate the utilities into your CI/CD automation pipelines.

Example CI/CD Architecture

In this example architecture we assume you have an existing CI/CD pipeline that can deploy Amazon Web Services services using CloudFormation and the Amazon Web Services SDKs.

Build

For the build stage of the architecture, the CI/CD pipeline initiates a Step Function workflow in the source environment which executes three Lambda functions, one for each resource type – asset models, assets, and dashboards. The Lambda functions can be run in parallel and use the export utilities to query the Amazon Web Services IoT SiteWise service API to generate the corresponding CloudFormation templates for the resources you wish to replicate. The Lambda function will then store the generated files in an Amazon S3 bucket for use during the deploy stage of the pipeline. For the S3 bucket, you can either use a common shared bucket across all of your Amazon Web Services environments or use S3 replication to automatically copy the files between separate buckets in each environment .

Deploy

In the deploy stage, the Amazon Web Services IoT SiteWise resources need to be created or modified in a specific order in the target environment, namely, asset models, assets, and dashboards. To do this, Amazon Web Services StepFunction workflow states are defined for each resource type and the workflow is configured to execute them in the proper order. Each workflow state will use Lambda function tasks that reference the corresponding CloudFormation template in S3. The resources first need to be created by the CI/CD pipeline, therefore the initial workflow deployment tasks will create the CloudFormation stacks.

Once the stacks are created, subsequent updates from the CI/CD pipeline will use the workflow and step functions to update those stacks which will modify and update the Amazon Web Services IoT SiteWise resources. The asset and dashboard states will wait for the previous state to finish deploying in CloudFormation before they start because they require those resources to exist before they can be created. Please see the architecture below for a visual representation.

For production workloads, customers can use CloudFormation change sets in their deployment pipeline and have a manual approval gate to verify the CloudFormation updates before they are made. Finally, if dashboards are part of your deployment pipeline, an Amazon Web Services IoT SiteWise Monitor Portal must be created beforehand in the target environment.

Conclusion

In this blog post we introduced the Amazon Web Services IoT SiteWise Tools for replicating Amazon Web Services IoT SiteWise resources between Amazon Web Services environments and showed an example architecture how they can be integrated into an automated deployment pipeline. We recognize that each organization has different requirements, procedures and tools when it comes to automating and deploying their IT infrastructure and applications. We designed the tools to be flexible to adapt the architecture for your own requirements and to be able to integrate them into your specific automation pipelines. We welcome improvements or additions to the utilities. If you have something to contribute back to the repository, feel free to submit a pull request in the repository for review.

About the Authors

Sebastian Salomon is a Sr IoT Data Architect with Amazon Web Services. He has 7+ years of experience in IoT architecture in different vertical like IIoT, Automotive, O&G, Smart Home, Smart City and Mining as well as data warehousing and big data platform. In the latest years he got focus in how to bring AI to IoT through scalable MLOps platforms. As a member of Amazon Web Services Professional Services, He works with customers of different scale and industries architecting and implementing a variety of end to end IoT solutions.

Ashok Padmanabhan is a Sr. IoT Data architect with Amazon Web Services Professional services focusing on Big Data Analytics & industry 4.0 solutions in Manufacturing domain.

Mihai Lucaciu , with over 16 years of experience, is a Senior IoT Data Architect at Amazon Web Services Professional Services, passionately helping customers with solution architectures, designs & implementations for various projects on industrial data, edge analytics and cloud services.

Tim Wilson is an IoT Enablement Specialist with Amazon Web Services’s Public Sector Partner organization. In this role Tim works with Amazon Web Services public sector partners to support their adoption and use of Amazon Web Services IoT services and solutions. He started at Amazon Web Services as a Solution Architect in 2012 when Amazon Web Services’s Public Sector business was relatively small. He has also held roles at Amazon Web Services managing an IoT prototyping lab and as a technical presenter in the Amazon Web Services Executive Briefing Center.