Every write to disk costs something, whether it’s wearing down an SSD, slowing I/O on a busy server, or draining battery on a laptop. One of the biggest offenders is logging. Between systemd-journald, web server access logs, and application-level logging, a default Debian 13 (Trixie) or Ubuntu 24.04 LTS system writes a lot to disk.… continue reading.
Every write to disk costs something, whether it’s wearing down an SSD, slowing I/O on a busy server, or draining battery on a laptop. One of the biggest offenders is logging. Between systemd-journald, web server access logs, and application-level logging, a default Debian 13 (Trixie) or Ubuntu 24.04 LTS system writes a lot to disk.… continue reading.
Thanks @hydn for this guide!
The only thing I was doing was clear journald logs once at week. Very useful post.
This is a really great article! I run a lot of “servers” at home, which is mostly just a collection of hand-me-down hardware from previous jobs or friends and it is all “aging”. I just had two SSDs fail on me in the past six months.
While waiting for new SSDs to come in, I went through a similar exercise of finding “how long can I make this failing SSD last” by eliminating as many disk writes/reads as possible. I’m bookmarking this article for future needs of the same nature.
+1 for remounting without a reboot!
+1 browser cache to tmpfs, insanely cool idea
+1 for the journald changes
I’ll be implementing several of these on devices where I don’t need the extra logging but I do enjoy having logs on my servers (I just love logs… I’m weird)
Thanks for writing this up!
@benowe1717 Thanks for the detailed feedback! Except for my work PC, I generally buy dirt-cheap NVMe and SATA SSDs. That, and microSD card failures (Raspberry Pi) have brought me to my knees a few times! ![]()