Muokkaa

Jaa

Intelligent apps using Azure Database for PostgreSQL

Azure App Service
Azure AI services
Azure Database for PostgreSQL
Azure Machine Learning
Power BI

Solution ideas

This article describes a solution idea. Your cloud architect can use this guidance to help visualize the major components for a typical implementation of this architecture. Use this article as a starting point to design a well-architected solution that aligns with your workload's specific requirements.

This article presents a solution for automating data analysis and visualization using artificial intelligence (AI). Core components in the solution are Azure Functions, Azure AI services, and Azure Database for PostgreSQL.

Architecture

Diagram that shows the dataflow of an intelligent application using Azure Database for PostgreSQL.

Download a Visio file of this architecture.

Dataflow

  1. An Azure function activity allows you to trigger an Azure Functions App in the Azure Data Factory pipeline. You create a linked service connection and use the linked service with an activity to specify the Azure function you want to execute.
  2. Data comes from multiple sources including Azure Storage and Azure Event Hubs for high-volume data. When the pipeline receives new data, it triggers the Azure Functions App.
  3. The Azure Functions App calls the Azure AI services API to analyze the data.
  4. The Azure AI services API returns the results of the analysis in JSON format to the Azure Functions App.
  5. The Azure Functions App stores the data and results from the Azure AI services API in Azure Database for PostgreSQL.
  6. Azure Machine Learning uses custom machine learning algorithms to provide further insights into the data.
    • If you're approaching the machine learning step with a no-code perspective, you can implement further text analytics operations on the data, like feature hashing, Word2Vector, and n-gram extraction.
    • If you prefer a code-first approach, you can run an open-source natural language processing (NLP) model as an experiment in Machine Learning studio.
  7. The PostgreSQL connector for Power BI makes it possible to explore human-interpretable insights in Power BI or a custom web application.

Components

  • Azure App Service provides a fully managed platform for quickly building, deploying, and scaling web apps and APIs.
  • Functions is an event-driven serverless compute platform. For information about how to use an activity to run a function as part of a Data Factory pipeline, see Azure Function activity in Azure Data Factory.
  • Event Hubs is a fully managed big data streaming platform.
  • Cognitive Services provides a suite of AI services and APIs that you can use to build cognitive intelligence into apps.
  • Azure Database for PostgreSQL is a fully managed relational database service. It provides high availability, elastic scaling, patching, and other management capabilities for PostgreSQL.
  • Azure Machine Learning is a cloud service that you can use to train, deploy, and automate machine learning models. The studio supports code-first and no-code approaches.
  • Power BI is a collection of software services and apps that display analytics information and help you derive insights from data.

Scenario details

The automated pipeline uses the following services to analyze the data:

  • Azure AI services uses AI for question answering, sentiment analysis, and text translation.
  • Azure Machine Learning supplies machine-learning tools for predictive analytics.

To store data and results, the solution uses Azure Database for PostgreSQL. The PostgreSQL database supports unstructured data, parallel queries, and declarative partitioning. This support makes Azure Database for PostgreSQL an effective choice for highly data-intensive AI and machine learning tasks.

The solution automates the delivery of the data analysis. A connector links Azure Database for MySQL with visualization tools like Power BI.

The architecture uses an Azure Functions App to ingest data from multiple data sources. It's a serverless solution that offers the following benefits:

  • Infrastructure maintenance: Azure Functions is a managed service that allows developers to focus on innovative work that delivers value to the business.
  • Scalability: Azure Functions provides compute resources on demand, so function instances scale as needed. As requests fall, resources and application instances drop off automatically.

Potential use cases

Azure Database for PostgreSQL is a cloud-based solution. As a result, this solution isn't recommended for mobile applications. It's more appropriate for downstream analysis in the following industries and others:

  • Transportation: Maintenance prediction
  • Finance: Risk assessment and fraud detection
  • E-commerce: Customer churn prediction and recommendation engines
  • Telecommunications: Performance optimization
  • Utilities: Outage prevention

Considerations

These considerations implement the pillars of the Azure Well-Architected Framework, which is a set of guiding tenets that can be used to improve the quality of a workload. For more information, see Microsoft Azure Well-Architected Framework.

  • For most features, the Azure AI Language API has a maximum size of 5120 characters for a single document. For all features, the maximum request size is 1 MB. For more information about data and rate limits, see Service limits for Azure Cognitive Service for Language.

  • In Azure Database for PostgreSQL, your ingress volume and velocity determine your selection of service and deployment mode. Two services are available:

    • Azure Database for PostgreSQL
    • Azure Cosmos DB for PostgreSQL, which was formerly known as Hyperscale (Citus) mode

    If you mine large workloads of customer opinions and reviews, use Azure Cosmos DB for PostgreSQL. Within Azure Database for PostgreSQL, two modes are available: single server and flexible server. To understand when to use each deployment mode, see What is Azure Database for PostgreSQL?.

  • Previous versions of this solution used the Azure AI services Text Analytics API. Azure AI Language now unifies three individual language services in Azure AI services: Text Analytics, QnA Maker, and Language Understanding (LUIS). You can easily migrate from the Text Analytics API to the Azure AI Language API. For instructions, see Migrate to the latest version of Azure Cognitive Service for Language.

Security

Security provides assurances against deliberate attacks and the abuse of your valuable data and systems. For more information, see Overview of the security pillar.

All data in Azure Database for PostgreSQL is automatically encrypted and backed up. You can configure Microsoft Defender for Cloud for further mitigation of threats. For more information, see Enable Microsoft Defender for open-source relational databases and respond to alerts.

DevOps

You can configure GitHub Actions to connect to Azure Database for PostgreSQL database by using its connection string and setting up a workflow. For more information, see Quickstart: Use GitHub Actions to connect to Azure PostgreSQL.

You can also automate your machine learning lifecycle by using Azure Pipelines. For information about how to implement an MLOps workflow and build a CI/CD pipeline for your project, see the GitHub repo MLOps with Azure ML.

Cost optimization

Cost optimization is about looking at ways to reduce unnecessary expenses and improve operational efficiencies. For more information, see Overview of the cost optimization pillar.

Azure AI Language offers various pricing tiers. The number of text records that you process affects your cost. For more information, see Cognitive Service for Language pricing.

Next steps