Posts

Showing posts with the label finds

Finds: "You Want Microservices, But Do You Really Need Them?" - A Docker Blog

I stumbled upon a great read from Docker titled “Do You Really Need Microservices?” , and it echoes a lot of what I personally lean toward in system design. While microservices are often treated as the industry’s holy grail, the article reinforces that monoliths still make a lot of sense especially for teams with limited resources. For me, understanding one coherent system, its purpose, and its domain is far more valuable than chasing architectural trends for the sake of it. A well-designed monolith is easier to manage, more predictable in both performance and cost, and avoids the operational overhead that microservices inevitably bring. Worth a read if you're evaluating architecture choices for your next project.

Finds: "How to do I clear system data on Mac OS?" - A Reddit Discussion

Mac, always running out of space, especially if you're a developer (Xcode, argh!). This isn’t a new find, but since I just ran through it again, I figured it’s worth posting here for my own reference and maybe it’ll help someone else squeeze out that last bit of disk space. Here you go:  How to do I clear system data on Mac OS?

Finds: "Microservices vs. Monoliths" - A Reddit Discussion

Here’s another interesting discussion I stumbled upon on Reddit. The debate between "Microservices vs. Monoliths" is a frequent one, but in my opinion, the context matters—both approaches have their own merits depending on the situation. I plan to write a more in-depth entry on this topic in the near future. Enjoy reading:  Microservices vs. Monoliths: Why Startups Are Getting "Nano-Services" All Wrong

Finds: "Software Estimation Is Hard" - A Reddit Discussion

Since I'm treating this blog like a post-it, I'd like to use it to reference interesting and relevant articles or discussions I come across on the web—something I can easily refer to again in the future. Here’s a good one:  Software Estimation Is Hard. Do It Anyway