Date: Mon, 19 Jun 1995 02:22:59 +0800 (WST) From: Peter Wemm <peter@haywire.DIALix.COM> To: hackers@freebsd.org Subject: How does the disk IO clustering work? Message-ID: <Pine.SV4.3.91.950610135321.14840F-100000@haywire.DIALix.COM>
next in thread | raw e-mail | index | archive | help
There's a mention that this was reimplemented in the "What's new in the VM system" doc that was posted here some time ago. What are the optimal ufs tunefs/mkfs/newfs parameters to get the best advantage from this? I remember reading the Sun white paper on their implementation some time ago. They needed special "rotdelay" and "maxcontig" parameters. Is there a difference in the layout from a 2.0R built filesystem? Is there anything to be gained by dumping everything to tape and rebuilding the partitions? (assuming there's a layout difference). The system that I'm using has a file system built like this: jhome # tunefs -p /dev/rsd0h tunefs: maximum contiguous block count: (-a) 1 tunefs: rotational delay between contiguous blocks: (-d) 4 ms tunefs: maximum blocks per file in a cylinder group: (-e) 1024 tunefs: minimum percentage of free space: (-m) 10% tunefs: optimization preference: (-o) time jhome # I dont know how it was built - but it would either have been made by 2.0R, or converted via fsck -c2 from a 1.x file system. (Julian?) Thoughts? -Peter
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.SV4.3.91.950610135321.14840F-100000>