Automated Deployment#

In a distributed edge network, edge nodes can consist of various device types and may be situated in remote areas, making them challenging to access and maintain. In the past, operators had to go to different locations to update the software on edge nodes manually.

In Edge Orchestrator, automated deployment lets you automatically deploy and update applications to clusters. There are two methods of automated deployment:

1. Metadata-based: Deploys applications to clusters based on labels that you create or are automatically created for you.

  1. Manual: Deploys applications to a list of clusters that you define.

Metadata-based Deployment#

In a metadata-based deployment, application deployment packages are assigned to clusters based on labels consisting of key-value pairs. Cluster metadata labels are either user-defined or automatically created based on location information. Deployment packages can be deployed to many clusters by specifying a label or set of labels they all share.

This method automatically scales. As new clusters matching a deployment package’s metadata labels are brought online, software packages are automatically deployed to them.

If an application deployment package is deployed with multiple metadata labels, the applications are only deployed to clusters that match all labels.

User-defined Metadata#

A wide range of deployment criteria, derived from business needs, can be satisfied by carefully assigning metadata labels to the edge clusters. Cluster labels are key-value pairs defined at cluster creation time. The key is called “Category” in the UI.

Depending on how you manage your deployments, you might define key-value pairs representing different customers, types of infrastructure, or environments. For example, if you manage a shared infrastructure for different customers and desire to deploy separate applications for each customer, you could define labels based on your customers:

  • Category: customer, Value: customer1

  • Category: customer, Value: customer2

Likewise, if you want to manage your application deployments using the blue/green deployment technique, you could define labels for each environment:

  • Category: environment, Value: blue

  • Category: environment, Value: green

With this set of metadata labels, for each customer you might deploy two copies of a deployment package: one to the blue environment and one to the green. In this case, it would be easy to upgrade the applications in customer2’s green deployment environment without affecting other customers or environments.

Location Information#

Each host is associated with a site and each site belongs to a region. You can create hierarchies of regions, allowing flexible infrastructure management. Labels associated with Sites and Regions enable mapping deployments to your physical infrastructure.

Each cluster will automatically be assigned metadata labels based on its position in the hierarchy. For example, if site1 belongs to region1, and region1 has parent region2, then clusters at site1 would get these labels:

  • Category: region, Value: region1

  • Category: region, Value: region2

You can also assign additional metadata labels to sites and regions that will be applied to the clusters. All location labels can be used to target application deployments in the same way as the user-defined metadata.

Enabling Automated Deployment#

To enable an automated deployment based on metadata, you must first do the following:

  • Provision Host, including adding location information (for the location-based labels).

  • Create Cluster, including adding user-defined metadata labels for the cluster.

  • Set up a Deployment, including adding the target labels for the deployment.

Once these elements are in place, Edge Orchestrator will automatically deploy software and updates to clusters matching the criteria. As new clusters matching the criteria are brought online, the software is automatically deployed to them.

Metadata Rules#

In the deployment process you must adhere to the metadata rules. Metadata enforces a lower case rule and does not accept spaces.

As Regions dynamically create metadata, they follow the same naming constraints as metadata.

Deployment and Site names may contain uppercase and lowercase characters, numbers, hyphens, slashes and spaces.

Manual Deployment#

Manual deployment is a simplified form of automated deployment. Instead of relying on cluster metadata labels, this method deploys applications to a custom list of clusters.

In this way, you can create a deployment target from clusters that do not share common criteria. Once created, the list of clusters cannot be modified.

See Set up a Deployment for information on associating a list of clusters with a software deployment.