ext2 for "in flight" data and ext3 for the "stable stuff"?
Posted Mar 20, 2003 23:42 UTC (Thu) by jzbiciak
(✭ supporter ✭
Parent article: Speeding up ext2
Sounds like these changes would tune ext2 more for transient data (stuff like /tmp, and scratch areas used by apps for works-in-progress), and ext3 more appropriate for "persistent" data.
Basically, anything that you could easily regenerate or mke2fs away across a reboot would live on whizzy-fast ext2 partitions.
I'm not saying that ext2 is unreliable. Rather, I'm saying you'd use ext3 wherever you don't want large fsck times and greater data protection guarantees, and ext2 on filesystems that are homes for high-bandwidth stuff that you don't mind losing if your machine crashes. You avoid the long fsck times for the ext2 filesystems across a crash by simply re-making them.
Squid caches, /tmp, web browser caches, GIMP caches, partially computed data sets, CDDA images, etc. all go in those 'scratch' areas on ext2 filesystems. In contrast, /home, /usr, /var, etc... all go in ext3.
Does that sound sensible?
to post comments)