The TechConf website allows attendees to register for an upcoming conference. Administrators can also view the list of attendees and notify all attendees via a personalized email message.
The application is currently working but the following pain points have triggered the need for migration to Azure:
- The web application is not scalable to handle user load at peak
- When the admin sends out notifications, it's currently taking a long time because it's looping through all attendees, resulting in some HTTP timeout exceptions
- The current architecture is not cost-effective
In this project, you are tasked to do the following:
- Migrate and deploy the pre-existing web app to an Azure App Service
- Migrate a PostgreSQL database backup to an Azure Postgres database instance
- Refactor the notification logic to an Azure Function via a service bus queue message
You will need to install the following locally:
- Create a Resource group
- Create an Azure Postgres Database single server
- Add a new database
techconfdb
- Allow all IPs to connect to database server
- Restore the database with the backup located in the data folder
- Add a new database
- Create a Service Bus resource with a
notificationqueue
that will be used to communicate between the web and the function- Open the web folder and update the following in the
config.py
filePOSTGRES_URL
POSTGRES_USER
POSTGRES_PW
POSTGRES_DB
SERVICE_BUS_CONNECTION_STRING
- Open the web folder and update the following in the
- Create App Service plan
- Create a storage account
- Deploy the web app
-
Create an Azure Function in the
function
folder that is triggered by the service bus queue created in Part 1.Note: Skeleton code has been provided in the README file located in the
function
folder. You will need to copy/paste this code into the__init.py__
file in thefunction
folder.- The Azure Function should do the following:
- Process the message which is the
notification_id
- Query the database using
psycopg2
library for the given notification to retrieve the subject and message - Query the database to retrieve a list of attendees (email and first name)
- Loop through each attendee and send a personalized subject message
- After the notification, update the notification status with the total number of attendees notified
- Process the message which is the
- The Azure Function should do the following:
-
Publish the Azure Function
- Refactor the post logic in
web/app/routes.py -> notification()
using servicebusqueue_client
:- The notification method on POST should save the notification object and queue the notification id for the function to pick it up
- Re-deploy the web app to publish changes
Complete a month cost analysis of each Azure resource to give an estimate total cost using the table below:
Azure Resource | Service Tier | Monthly Cost |
---|---|---|
Azure Postgres Database | Basic | $24.82 |
Azure Service Bus | Basic | $0.05 |
Azure Function App | Consumption 128 MB memory | Free |
Azure App Service | Basic (B1) | $12.41 |
Azure Storage Account | General purpose (V2) | $82.25 |
The project cosisnts of web app configered with a database, and a function to send notifications to attendees. The new implementation of the project consists of a migration to an Azure App Service and Azure Postgres database instance which improves cost efficiency, and an azure function that is triggered by a service bus queue.
The built in web application architecture is not scalable to handle clients' requests such as the admin filling the notification form and sending the request, in the meantime a lot of requests can be submitted and the application could not handle all the load which results in HTTP timeout. Azure offers services available for long-running tasks, in this case, we used Azure Function via a service bus queue message to handle loads of requests, which leads to making the performance of the application higher as well as making it scalable and cost-effective Through the migration to a microservice architecture and refactoring of the notification logic to an Azure Function via a service bus queue message, the different components of the web application are decoupled. This makes it more scalable and sending out notifications.