3 ms·

author here. so, this was an object lesson in not running every single one of your services on a single ec2 micro instance. sorry about that.

• latch 15y ago
  Don't know anything about typo, but an alternative to throwing more hardware at it could be to optimize your site. My story (redis/mongodb) peaked at the same time as yours, and my linode 1024 ($40/m), which also houses about 7 different projects, didn't go past 5% cpu usage. Going with Disqus lets you outsource the only dynamic part of your blog post. This lets you set a long cache header, or at least to serve it from cached memory. Even if you don't outsource your comments, I bet caching the page for 2-3 minutes in memory would solve most of your problems. Also, cache headers on your other assets wouldn't hurt. Nor would moving away from Apache. Sounds like a good weekend project :)
  • mwotton 15y ago
    turns out i had typo configured to serve everything through rails, as well as having 10 instances running. not a happy combination.