The conclusion is a bit frustrating, even stating “Without throwing additional money at the problem…” because taking the time to learn enough and then to execute such tests surely cost significantly more than simply upgrading the server from what sounds like Linode’s lowest offering to something slightly more powerful. This article was clearly written to make a point, not to actually be a good business move: which is fine, just it’s not what younger engineers should be emulating.
I’ve seen lots of things like this lately in articles and youtube videos. Here’s a problem, here’s how I benchmarked it, here’s what I decided to do. But never do they explain that the economics of doing it this way are completely wrong for most business situations (it’s fine for hobby work, however).
Optimizing the solution you already have, or throwing a small amount of money at a problem, is often the right move for a business. Doubling the RAM and vCPU of the Linode VPS would cost a total of $30/month, for a blog which appears to be a big part of Twilio’s marketing operation. Linode don’t even sell new customers the size of VPS as the author uses. Or reorganizing how the data is stored within the database, which doesn’t even seem to be touched on, but in my understanding can have a significant impact on query performance.
Very true – doubling the size of a VPS is a great way to get more performance fast, and for most business operations, cost is not a big consideration. The business considerations are left out of many discussions and is a useful perspective – thanks for contributing.