Arjay McCandless wouldn't ship his fastest server experiments
Community optimizations squeeze far more from a $12 machine, but maintainability gets its own turn in the spotlight.
Why this story? The channel's weekly summary: what went up this week, what is drawing views now, and where the channel stands among similar channels.
I Asked Expert Programmers to Max Out a $12 Server
Things we can learn from the video
Press to jump
- 1:16Submissions must run matching API endpoints, verify JWTs and stay within latency and error limits during sustained testing.
- 2:42The traffic generator runs out of memory, requiring four virtual machines to put enough pressure on the server.
- 15:47Rust and C++ reach the test's practical connection limit; Rust has lower latency and more CPU headroom.
- 20:00Faster implementations reuse prepared queries, batch writes and sometimes replace frameworks with custom servers.
- 21:11Extreme throughput comes with maintainability, security and portability compromises that Arjay would reject for large workplace applications.
- 21:54Grouped writes can share a failure, while a single database connection can make other requests wait.
Getting the server to handle more traffic is only half the lesson. Near the end of "I Asked Expert Programmers to Max Out a $12 Server," Arjay McCandless delivers the qualification that makes the experiment useful beyond its leaderboard: "I probably wouldn't ship most of these implementations or approve these PRs for a large scale application at work."
The community has pushed his little machine to 64,000 concurrent users, more than four times his previous best. Getting there sometimes means abandoning standard frameworks, writing custom servers and parsers, or accepting code that is harder to extend. Arjay gives those achievements their due, then separates winning a tightly defined benchmark from building something colleagues can safely maintain.
That distinction has supplied the biggest draw among his five uploads this week. The community challenge has reached roughly 450,000 views. In the comments, viewers keep returning to how often optimizations involve C libraries; others single out the work on SQLite itself. One thoughtful thread questions the failure-handling complications of grouping database writes together, precisely the sort of trade-off Arjay addresses in his closing analysis.
The surrounding uploads make this feel like an ongoing investigation rather than an isolated test. "I Ran 4 Databases on the Same $12 Server" changes the database while keeping the Go API and data consistent. "Serverless API Recipe" adds an AWS system-design lesson, alongside a Node-and-Bun speed comparison. Meanwhile, the earlier programming-language experiment remains the channel's next-busiest video behind the community follow-up.
Arjay added 6,000 subscribers over seven days, and daily viewing is now about triple its usual pace. Nearby in subscriber size, The Tech Report and Microsoft Helps are having slower-than-usual viewing days. For Arjay, the inexpensive server has become a productive teaching tool: not just how to make software faster, but what that speed can cost.
Published on Trari
Where next?
What viewers say
I love how most languages speed up was: "Use a c library".
@kolumdiumSee it on YouTubeOptimizing the SQLite binary itself is insanely impressive. Love the dedication.
@sarthakgothalyan8952See it on YouTubeC absolutely chillin' with those 39mb of ram
@drdestenSee it on YouTube
3 more comments
Language: support 40k upwards users
Looks underneath: C@davixpixie243See it on YouTubeThe biggest lesson is that basically any language can reach the top level of performanc by stripping away nearly everything
@pretro6136See it on YouTube"I think there could be a little more performance hiding in this language"
Yeah. It's C. If you can't squeeze out more performance there you pretty much have to start writing assembly.@emilemil1See it on YouTube
Public comments from the video's page on YouTube
The videos drawing views on Arjay McCandless right now
I Ran 4 Databases on the Same $12 Server164,408 views · 2026-10-09
Serverless API Recipe76,718 views · 2026-10-09
Node vs Bun (which is faster?)68,746 views · 2026-10-10
Why is PostgreSQL faster than SQLite?49,084 views · 2026-10-12
Which Programming Language Can Handle the Most Users on a $12 Server?718,450 views
- Channel
- Arjay McCandless
- Subscribers
- 124,000 ▲ 6,000 this week
- Channel views, last 7 days
- 1,642,936
- Videos published this week
- 5
- Video published
- October 13, 2026
- Video views so far
- 448,026
- Views on its first day
- 384,799
The numbers
Here is how Arjay's server-testing week looks in views and subscribers.
💡 Test your knowledge
Which language powers the API in Arjay's four-database comparison?
Arjay keeps the same Go API while testing Postgres, MySQL, MariaDB and SQLite.


