Two Scoops Study Session #19

Chapter 29 covered separating common functions or classes into utils.py to improve reusability and testability, and noted that django.utils is mainly intended for Django's internal use, so its stabilization status should be checked before use.

Regarding exception handling, it was emphasized that Django provides various exceptions, and that while JSON, Python, YAML, and XML data can be handled with serialization and deserialization tools, they are vulnerable to changes in data structure. Therefore, the guideline was to use them only for simple data, and to always validate with Django forms before saving.

Chapter 30 viewed PaaS as a useful deployment method for small projects, while emphasizing the importance of choosing options that are not tied to a specific service. The selection criteria mentioned were legal constraints, pricing, uptime, support staff, ease of scaling up, level of documentation, geographic requirements, and the company's stability.

As deployment principles, it was emphasized that matching the development and production environments, deployment automation, verification through a staging server, and preparing backup and rollback are key. This led to the message that deployment becomes safe only with automation and prior preparation.

In the advanced session, the concept and necessity of container orchestration were covered, and DC/OS, one of the tools, was introduced along with hands-on practice. DC/OS combines the Marathon Framework on top of the Apache Mesos distributed systems kernel, with the goal of automating and optimizing the deployment of multiple containers.

In the hands-on practice, it was particularly impressive that resource allocation and logging could be intuitively checked on the dashboard, and package management could be done with a button click. Finally, along with the sadness that the study session had ended, thanks were shared while looking back on the summaries and advanced materials from the past 5 months.