1. Enhanced Monitoring Workflows for AI Applications via AWS SSM Contacts


    To set up enhanced monitoring workflows for AI applications using AWS Systems Manager (SSM) Contacts, you'd typically want to define contacts and escalation plans that notify the appropriate teams when certain events or conditions are detected in your AI applications. Using AWS SSM Contacts, you can manage on-call schedules and escalation chains to automate and enhance the response to operational incidents.

    In Pulumi, this can be done by defining resources for contacts and escalation plans (also known as contact plans). You will need to create an SSM Contact resource representing each individual or team to be contacted in case of an incident, and an SSM Plan resource to specify the escalation path and rules.

    The following Pulumi program is an example of how to create these resources in Python:

    1. Contact - Represents the individual or team to be contacted. Here, you specify the name and type of the contact.
    2. Plan - Defines the escalation plan, which includes the stages of how and when different contacts are notified.

    Let me walk you through the Pulumi program that sets up a basic monitoring workflow with a contact and an escalation plan.

    import pulumi import pulumi_aws as aws # Create an SSM Contact resource. This contact represents an individual or team. contact = aws.ssmcontacts.Contact("aiAppMonitoringContact", alias="ai-team-contact", display_name="AI Application Monitoring Team", type="PERSONAL", # or 'ESCALATION' for teams ) # Create an SSM Contact Channel. This is the method through which the contact will be notified (e.g., SMS, Email). contact_channel = aws.ssmcontacts.ContactChannel("aiAppMonitoringContactChannel", contact_id=contact.id, delivery_address={ "simple_address": "example@example.com" # The email address for the contact. }, type="EMAIL", # The type of contact channel ('EMAIL', 'SMS', 'VOICE') ) # Create an SSM Plan resource that defines a specific escalation plan for the contact. plan = aws.ssmcontacts.Plan("aiAppMonitoringEscalationPlan", stages=[ { "duration_in_minutes": 15, "targets": [ { "channel_target_info": { "contact_channel_id": contact_channel.id, "retry_interval_in_minutes": 5, }, "contact_target_info": { "contact_id": contact.id, "is_essential": True, }, }, ], }, # Additional stages can be added here to escalate the alert to other contacts or teams. ], ) # Export the contact and plan information pulumi.export("contact_alias", contact.alias) pulumi.export("contact_channel_type", contact_channel.type) pulumi.export("escalation_plan_name", plan.name)

    In this program, we create a contact for AI application monitoring using aws.ssmcontacts.Contact. We specify that this is a personal contact, though you can also specify an escalation plan for a team using the ESCALATION contact type.

    Next, we create a contact channel using aws.ssmcontacts.ContactChannel, which in this example sends notifications over email to the specified address when the AI application monitoring triggers an alert.

    Finally, we create an escalation plan using aws.ssmcontacts.Plan, which outlines the actions to be taken when an incident occurs—the example here only has one stage, but you could add more stages to escalate to other contacts or teams.

    This is a basic setup and can be expanded with more complex workflows, multiple contacts, additional contact channels (e.g., SMS, voice), and various escalation patterns to match your application's operational requirements.

    Keep in mind this is a starting point for setting up monitoring workflows. You would likely need to integrate this configuration with cloudwatch or other monitoring services to detect and respond to incidents in your AI applications.