Million Dollar Lines of Code: an Engineering Perspective on Cloud Cost Optimization

InfoQ Dev Summit Boston, a two-day conference of actionable advice from senior software developers hosted by InfoQ, will take place on June 24-25, 2024 Boston, Massachusetts.
Deep-dive into 20+ talks from senior software developers over 2 days with parallel breakout sessions. Clarify your immediate dev priorities and get practical advice to make development decisions easier and less risky.
Register now: bit.ly/47tNEWv
---------------------------------------------------------------------------------------------------------------------
Video with transcript included on InfoQ: bit.ly/3QGY6o7
Erik Peterson discusses the right timing and approach for engineering cost optimization and how to use cost efficiency metrics as powerful constraints that drive innovation, and engineer profit.
#CloudComputing #CostOptimization #InfoQ
---------------------------------------------------------------------------------------------------------------------
Follow InfoQ:
- Mastodon: techhub.social/@infoq
- Twitter: / infoq
- LinkedIn: / infoq
- Facebook: / infoqdotcom
- Instagram: @infoqdotcom

[@k_tully]

Excellent work.

[@canismajorisvy]

thanks for this talk!

[@borisjoffe]

dhh said they spent tons of time optimizing cloud costs and it was still overpriced - that's why the switched to their own servers.
The fact that it's so easy to make mistakes costing millions is a serious problem with the cloud.
If you look at the basic economics, cloud companies have to make a profit, so they're going to charge more for a server. And the layers of virtualization hurt performance compared to a bare server. So for many companies with steady traffic, on-premise is simpler and cheaper. For services with very variable demand, the cloud can be more cost effective

[@mihaisagungureanu3070]

It's a new way of working. Don't forget you CAN set budgets and stop the phenomenon when it hits the limits.
The cloud sliced and diced the resources to the single "write" on the storage. That's what you pay for. You need to think on the same abstraction level when you code.

[@borisjoffe]

​@@mihaisagungureanu3070 Setting budgets is definitely good, but for most companies and most applications, using ultra-specific cloud tools, Kubernetes, Docker, cloud functions, etc is just overkill and increases vendor lock-in.
A cheap fixed-price per month server with nginx, your app backend, and a database is enough for most companies

[@FlaviusAspra]

So what's the single character we're talking about at 20:00?

[@piengeng]

typo on the company's url?

[@AntonioDoesMetal]

It was a typo in the url for the word “company” as the other person mentioned. This means that when the function download_update_metadata ran its response code was never 200 because it was a bad URL. So it fell back to doing a full download and file based hash every single time in the wake condition