Edge Developer Toolbox Developer Guide
Bring Your Own Application Artifacts
Applications designed for the hybrid scenario may include containers, source code, Dockerfiles, or deployment resources such as Helm chart. To import one or more of these artifacts, click Develop AI Application -> Bring Your Own Application. Alternatively, you can import many of these resources through the VS Code IDE (See Quick Start with a Reference Implementation in VS Code).
Click the appropriate resource to import and configure them.
Container from Registry
Use this section to import existing Linux* containers from public or private container registries.
Popular registries supported are Docker Hub* service, Amazon Elastic Container Registry*, Azure Container Registry*, Google Artifact Registry*, and Red Hat* Quay.io Container Registry. Always import containers from sources you trust.
Importing from private registries requires entering the credentials to the repository.
Example: Import a container from Docker Hub. For this exercise, we will import a simple nginx container.
Repo URL:
docker.io/nginx:stable-perl
Application Name:
nginx
Click Import.
Once imported, containers can be found under My Workspace > Application > Container and can now be used to launch and benchmark on Intel hardware or within other applications.
Supported Actions:
Configure runtime parameters. See the Configure Containers section
Launch containers on Intel hardware and benchmark. See the Benchmark Hardware section.
Publish to Repo enables pushing the container image to a repository of choice.
Configure Containers
Each container in an application can be configured for additional runtime parameters.
Click Actions -> Application Configuration
Following are the configurable parameters:
Port: To expose ports for multi-container communication.
Enable Routes (Toggle): To access your service through a URL.
Labels: To create labels for keeping track of custom configurations of an imported container across projects.
Entry Point: To override the default entry point.
Output Mount Point: To retrieve the result data from the container to be previewed later in your filesystem.
Volume Binder > Filesystem Path: To enter path(s) (maximum: 5 paths) on the Intel® Developer Cloud for the edge filesystem.
Volume Binder > Input Mount Point: To enter path(s) (maximum: 5) inside the container.
Dependency: In multi-container scenarios, to select the other dependent containers that must run before this container begins to run.
Configuration Parameters: To pass runtime container environment variables for dynamic container configurations, for example, -e VAR1=VALUE1 -e VAR2=VALUE2. For alternate entry point parameters, also referred to as command args, use Docker* Compose files or Helm* charts instead.
Check for valid configurations such as mount points to avoid unexpected launch failures.
Ensure path names do not contain spaces.
Git Repository
Import source code from public and private repositories by providing the repository URL, application name and credentials (for private repositories).
Java, Python, PHP, NodeJS, and Go are supported. Additionally, you can import Dockerfiles.
Once imported, the source code is available under My Workspace > Application > Source Code.
Supported Actions:
Configure runtime parameters. See the Configure Containers section
View Build Logs and troubleshoot applications
Launch containers on Intel hardware and benchmark. See the Benchmark Hardware section.
Publish completed applications to a repository of choice.
Helm Chart
You can import Helm* charts from existing repositories such as Bitnami* application and Artifact Hub application, or from a source code repository.
Alternatively, you can also import Helm charts from the Edge Developer Toolbox file system.
Enter the Repo URL, Repo Name, and a Chart Name from the repository to begin the import process.
Repository Notes
For code repositories, the portal will look for an index.yaml file that references one or more archived charts ending with .tgz.
You can create packaged charts from a chart directory with the Helm* package PATH_TO_CHART_DIR and generate the index file with the Helm repository index ahead of time.
For repositories such as the Bitnami* application, use paths that are generally used with the Helm repository add command. Example: https://charts.bitnami.com/bitnami.
For source code repositories, enter the raw GitHub* path format. To retrieve the raw GitHub file path, open the index.yaml file in your browser and select to open Raw file contents. Ensure the filename only includes the repository name and branch. Example: https://raw.githubusercontent.com/intel/DevCloudContent-helm/main
The Edge Developer Toolbox modifies the imported Helm chart if necessary to enable successful execution on the platform.
Successfully imported Helm packages can be found under My Workspace > Application > Helm Chart.
Supported Actions:
Launch containers on Intel hardware and benchmark. See the Benchmark Hardware section.
Publish completed applications to a repository of choice.
Publish to Edge Orchestrator (coming soon)
Configure Helm import, reimport and repackage for deployment.
Docker Compose
You can easily run multi-container applications on various Intel platforms by importing Docker* Compose YAML files hosted on external GitHub* repositories.
Provide the Git repo URL, private repo credentials (if needed) and a unique Resource Name to import the Docker Compose file. The Edge Developer Toolbox modifies the resource if necessary to enable successful execution on the platform.
Successfully imported Docker Compose files are available under My Workspace -> Application -> Docker Compose.
Common Docker Compose configurations such as depends_on, ports and build are also supported.
Make sure you place a single Docker Compose file in the root of the repository because the container-playground portal will look for the first available docker-compose.yml in your Git* repository.
Supported Actions:
Launch containers on Intel hardware and benchmark. See the Benchmark Hardware section.
Command Line Import
The Edge Developer Toolbox provides a Command Line Interface that allows source code pull, compilation and build options.
Launch a Terminal by clicking on + and Terminal
A new terminal with Java* 11, Python* 3.8.2 kernel, Maven*, Git* and Buildah* tools preinstalled for developers’ convenience is now ready for use.
Images built using the CLI can be found in the user’s My Workspace -> Application.
AI Models
Custom trained AI models are an integral part of developing AI solutions. While users can import default AI models from the Open Model Zoo, or custom models for further optimization using the OpenVINO™ toolkit (See theModel Optimization section), the Edge Developer Toolbox also allows developers to import custom AI models from Azure* or Amazon* S3 buckets.
Click My Workspace -> Filesystem to access the Azure or Amazon S3 connectors.
Provide the required credentials to your account to proceed with downloading AI models and other solution artifacts to the local file system.