Microsoft Power Platform Architecture

Source: My notes from Microsoft session led by Sravani Seethi on 2023-02-23

Agenda

• Power Platform pillars and solutions
• Design:
  • Selection of applications, automation, data storage
  • Applications classification

Pillars and Solutions

• Pillars: Power Pages, Power Apps, Power Automate, Power BI, Power Virtual Agents
• Tools: Connectors, Dataverse, Power Fx, AI Builder
• Pro developers can also use all Azure and data services

Solutions

• Packages all Power Platform pieces into one place, packaged as a unit

• Types:

  • Managed
    • Solution is locked, cannot be edited
    • Used for deployments
  • Unmanaged
    • Open folder so more things can be added like flows, Power Platform assets

• Solutions help with migration, collaboration, consistency

• Lifecycle: Make changes, export, install, patch, roll up patches and start changes again

Designing

Application Selection

1. Canvas vs Model-driven Apps

   • Canvas:
     • For user experience
     • Visually appealing
     • Mobile and table apps
     • Can use all connectors
     • Limited capabilities are included in Office 365 licenses
     • Security and access is managed on data source
     • Can use entities in Dataverse
   • Canvas pros and cons
     • Good for business and pro developers
     • Personal to enterprise use
     • Data source flexibility
     • Custom UX
     • Gaps and alternatives:
       • Reporting - use PowerBI
       • Performance - use multiple canvas/model apps
       • Time to build - use reusable components, containers, generate apps from images/figma file/API
       • Delegation - use with data model design
       • Limited UX controls, though purposely limited for accessibility - use custom Power Apps Custom Framework (PCF) with Typescript
       • External non-guest users - use with Power Pages
   • Model-driven:
     • Start with Dataverse database
     • App building using set components
     • Web and mobile apps
     • Tied to Dataverse tables
     • No capabilities within Office 365 licenses
     • Security is role based and row level security to data
     • Can use business process flows, code add-ins. real time work flows
   • Model-drive pros and cons
     • Business and pro developers
     • Out of box controls, rules and processes
     • Data ETL tools
     • Built in Excel and Word generation
     • Hybrid canvas/model
     • Needs Dataverse
     • Gaps and alternatives
       • Large data volumes - virtual entities
       • Limited UI modification - custom controls, canvas app, JS, PCF
       • Limited to Dataverse - use canvas app
       • External non-guest users - Power Pages

2. Selection Guide

   • Non-dataverse, use canvas app
   • Complex data, use model driven
   • Minimal licensing, use canvas app

Power Virtual Agents

• Low code
  • SaaS hosted low/no-code
  • Using Azure bot service
  • Voice channels
• Pro code
  • Open source framework
  • Build bots from code
  • Composer offers visual authoring

Automation Selection

1. Power automate and custom connectors

   • Flows
     • Like automations, work flows, validations
     • Alternatives to meet gaps:
       • Logic Apps
       • Dataflows
   • Custom connectors
     • For pro developers
     • Reuse API, use APIs
     • Alternatives to meet gaps:
       • Azure API, functions

2. Power Automate Flows

   Use different triggers

   • Cloud flows:
     • Uses different triggers
     • Automated, instant, scheduled
     • Use case: process automation and batch processing
   • Desktop
     • Use case: UI and desktop automation
   • Business process flow (BPF)
     • Use case: Business process automation
     • Works with Dataverse
   • Can be limited by licensing for calls. Consider logic apps for high volume of calls

Data Selection

1. Data Flows

   • ETL tool to extract data, tranform data, and load to destination like Dataverse
   • Use DAX language

2. Scalable Data

   Evaluate security, capacity, complexity to choose data sources. Common ones are:

   • SharePoint
   • Dataverse
   • On premise data
   • Cloud storage

   1. Sample Use Cases

      • Simple team app: SharePoint or Dataverse with canvas app
      • Small organization app: Dataverse or SharePoint with smaller data with canvas or model app
      • Complex business application: Dataverse, SQL with canvas/model app
      • Enterprise, large applications: SQL with canvas/model app

3. Power Apps Delegation for Data

   • Power Apps will delegate data processing to the data source rather than doing processing locally
   • Power Apps is limited to 500-2000+ records to local device / browser

Solution Complexity Drivers

Drivers:

• Number of:
  • Users and stakeholders
  • Customizations
  • Business rules
• Organization(s)
• Requirements maturity
• Data volume
• Testing needs
• Deployment
• Support model (app owner to large technical team)
• Delivery model like citizen developer, pro developer, to team development
• Development cycle
• Time allowed for MVP
• Change frequency
• Return on investment / “payback”
• Suitability to Power Apps use cases
• App criticality to business based on above drivers