Date: Thu, 3 Apr 2003 12:17:48 +0100 (BST) From: Jan Grant <Jan.Grant@bristol.ac.uk> To: Heinrich Rebehn <rebehn@ant.uni-bremen.de> Cc: questions@freebsd.org Subject: Re: newfs parameters for a 500 GB volume ? Message-ID: <Pine.GSO.4.44.0304031216580.22089-100000@mail.ilrt.bris.ac.uk> In-Reply-To: <3E8C12C8.5050503@ant.uni-bremen.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 3 Apr 2003, Heinrich Rebehn wrote: > Jan Grant wrote: > > >On Thu, 3 Apr 2003, Heinrich Rebehn wrote: > > > > > > > >>Thank you for your reply. I have reread the manpage and would go for: > >>- blocksize => 64 kB > >>- fragsize => 64/8 => 8 kB > >>- bytes/inode => 16 kB > >> > >>Are these values reasonable or should i use higher values (because the > >>disk is so large?) > >> > >> > > > >I'd turn bytes per inode up to 200k or so - there's a paragraph in "man > >7 tuning" on this. > > > >If the application is critical then benchmarking "real use" as closely > >as possible will enable you to find optimal sizes for stripes, and is > >probably time well spent. > > > > > > > > > Oh, i had forgotten the tuning(7) manpage. Neiter newfs nore tunefs > contain a refrence to it! > Having read it, i am not so sure about the blocksize any more.I > t says that going higher that 16K can make the fs slower. Does this > still hold true? As far as that's concerned, your best bet would be to benchmark the application you're intending on using and see how it behaves. -- jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/ Tel +44(0)117 9287088 Fax +44 (0)117 9287112 http://ioctl.org/jan/ It's a sad fact that the word "semantics" seems to have lost all meaning.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.GSO.4.44.0304031216580.22089-100000>