Developing software for the Internet of Things (IoT) comes with its own set of challenges. Security, privacy, and unified standards are a few key issues. In addition, each IoT product is comprised of at least 3 separate application components: the software embedded in the device, the backend big-data service, and the mobile application for the end user’s controls. Each component is developed by a different team, using different technologies and practices, and deployed to a different stack/target – this makes the integration of these separate pipelines and the coordination of software updates for IoT more problematic. How do you coordinate the diverse moving parts that must come together when your IoT product is updated?
Getting IoT to Flow
Electric Cloud helps our customers solve software delivery problems at a very large scale. Our integrated, end-to-end, DevOps platform – ElectricFlow – has proven to be a natural fit for orchestrating the complex pipelines that are common with IoT.
ElectricFlow 6.0 introduces Pipelines to allow you to automate end-to-end software development and delivery processes, enabling Application Release Automation (ARA) and Continuous Delivery (CD). You can easily manage the software delivery of one or more applications using pipelines.
A pipeline is a series of high-level re-usable steps that run automated processes. Pipelines consist of one or more stages. Each stage has a stage plan with one or moretasks, an entry gate, and an exit gate.
Tasks are used to run automated build, test, and deployment processes. The tasks within a stage plan are run in sequential order.
An entry gate and an exit gate exist for each stage. A gate may consist of one or more automatic or manual approvals. When the tasks for a stage are completed and an exit gate requires approval before going to the next stage, the approvers are notified by email.
Pipelines provide several benefits:
- Orchestration – determine who (or what) is working on the release process, what part they are doing, what is the status of that part, and what is the next step.
- Visibility – determine how the software is performing, whether it is ready for beta or production, where there are performance issues, and what versions are available within the release process.
- Control – determine whether the software is ready for integration, beta, preproduction, or production and whether the software passes the quality criteria at stage gates and has been approved by the appropriate users or groups.
Pipelines in Action
Let’s walk through a real use case to see how ElectricFlow is used to coordinate an IoT release.
In our example, there are three teams that work on the major components of a car-based IoT product:
- The embedded team develops the software deployed to electric cars. This software collects vehicle-specific information (battery charge, MPGe, etc.) and uploads it to the data center.
- The backend data team develops the software deployed to the data center. This software collects, analyzes, and provides visualization for the data uploaded by the vehicles.
- The mobile team develops the app deployed to the app store. This mobile app shows car owners real-time information about their vehicle, such as battery utilization.
For the demo, we assembled the following pieces running “v1” of the product:
The vehicle – the embedded software in the vehicle is simulated by a Raspberry Pi with a Blink(1)light indicating the battery charge and pushing information to the DB
The data backend– simulated by a database running in AWS EC2, and a Dashing dashboard to visualize the data
Working in Isolation
Each team uses the appropriate automation mechanism to build, test, and deploy their software in isolation from the other teams.
The vehicle and backend data teams:
These teams use ElectricFlow to:
- Model their applications’ tiers and components
- Define their processes used to build, test, and deploy their applications
- Define and run pipelines to execute these processes across the various stages of development.
The backend data application model
The vehicle deployment process
The backend data pipeline
The mobile application team:
Using Ship.IO, the mobile team builds, tests, and deploys their solution to various test devices as well as to the App Stores.
The mobile app jobs – build triggered automatically with each commit to Git.
The mobile app build/test/deploy process
Bringing It All Together
While lower environment development and testing is done using isolated pipelines, all three applications must converge, and be tested, before the final push to production. ElectricFlow manages this Release Pipeline.
The Release Pipeline stages and gates in ElectricFlow
For the purpose of the demonstration, we update each one of the three components comprising the IoT service to form a “v2” of our product. Once all three updates have passed their pipelines, we are ready to stage our coordinated release. To do that, we run the Release Pipeline.
A run-time instance of the Release Pipeline
The applications use snapshots to deploy the exact same bits and processes in each stage of both the individual team pipelines and the coordinated Release Pipeline. The exact same applications are deployed and tested from Development through to Production. This ensures repeatability and consistency, and greatly reduces the risk of failures when deploying to Production. When the applications converge, a successful deployment to the Staging environment is followed by a successful deployment to Production.
A successful Release pipeline
The ElectricFlow environment inventory shows the version of each component that is currently deployed. This visibility is powerful both when debugging failures and providing data for audit reports.
The inventory for a Staging environment
This example represents the moving parts that are typical of an IoT service. Coordinating the software delivery for these moving parts poses a challenge, particularly when cross-teams integration is required, and especially at scale. While manual coordination is typically used to address this, manual tasks are slow and extremely error-prone. Automation is the key to high-quality software being delivered at a rapid pace.
In the above demonstration, we see how ElectricFlow models an entire end-to-end IoT delivery lifecycle. Pipelines orchestrate the development and subsequent convergence of multiple applications – from commit through test and deployment to production. The simple and intuitive UI, along with the scalable automation of CI and deployment processes, makes ElectricFlow a natural solution for IoT software delivery challenges.