gino kaleb SYS ADMIN Blog
Measuring recovery, not perfect TPS
My load tests failed every time TPS dropped below 19.5, and that told me nothing. I redesigned them to measure what matters: how much the server degrades and how long it takes to recover.
Read post →An anti-lag that manages instead of deleting
Classic anti-lag systems free up resources by deleting things at random. Aurora Optimizer splits the problem in three: one component observes, another decides based on context, and another applies the fix precisely.
Read post →What a rules engine taught me about AI
My decision engine had one key rule: when a request involved a health issue, don't decide, pass the case to a person. Working with modern AI made me value that boundary even more.
Read post →A rules engine for academic requests
I automated request approvals with a layered rules engine. The most interesting part was deciding which cases the system resolves and which ones go to a human.
Read post →A simulated backend in the browser
To keep the frontend moving, I replicated the backend logic in JavaScript and used localStorage as the database. That let me test the whole system without a single server.
Read post →Choosing hosting for a static site
I needed free, fast hosting with automatic deploys on every push. I compared the options and went with Cloudflare Pages; here's why.
Read post →Inside IT-Support-Scripts
How I organized the toolkit: standalone scripts per task, an orchestrator with a menu, central error handling, and HTML reports anyone can read.
Read post →