09-10-2024: Meeting Agenda

[hansaskov]

This is the meeting agenda for our master thesis. Feel free to write comments for the topics for the agenda.

The meeting will be from 16:00-16:30

[Sebastian689]

Topic 1 Authentication and Authorization

1. IAM
2. IdP

[hansaskov]

Topic 2: Report structure.

We were thinking of spending some more time writing on the report. We would like to hear your opinion in regards to the general structure and what to include within the methodoligy section.

Here it is the read only overleaf document https://www.overleaf.com/read/wytmgfxxynwq#a37d46

[hansaskov]

Topic 3: CI/CD Testing on Preview Branch vs Pull Requests

Current Architecture

Let's say we have the following backend architecture, where we have a reverse proxy, a set of stateless services like the ingestion service and backend, and then a stateful service like PostgreSQL.

Now let's simplify it to only consist of the reverse proxy, the stateless and stateful services.

Deployment Options for Pull Requests

My question is: If we were to have preview deployments for pull request, how would that deployment look like?

Option 1: Isolated Database for Each Pull Request

Should we create a new isolated database for each pull request?

In this diagram:

• We have the main branch (production deployment) at the top
• Below, we have identical deployments with isolated databases for each pull request

Option 2: Shared Preview Database

Or should we only deploy the stateless services and reuse a testing database?

In this example:

• We have the production deployment
• A preview deployment where pull requests will use the database from the preview deployment instead of creating their own

The benefit of this approach is that it would require fewer resources.

Discussion

What are the pros and cons of each approach? Which would be more suitable for our CI/CD pipeline?