CHAA006%vaxb.rhbnc.ac.uk@CS.UCL.AC.UK (09/26/86)
Cluster factors: I think the most significant part of your message is in the last paragraph --- >>> Except for a few big indexed files (which are designed with FDL etc.), >>> almost all the files on our VAXen are simple variable length record >>> sequential files, mostly small, but with enough larger ones to make the >>> disk fragmentation we are currently seeing a nuisance. When deciding cluster factors for the Bedford College system, I asked a statistically-literate friend to perform a simple statistical analysis of the distribution of file sizes on our existing discs. It shewed without doubt that the modal filesize was one block, and therefore that setting the cluster factor to anything greater than 1 would waste an enormous amount of potential disc space. It sounds as if the same situation may obtain on your system. ** Phil. Philip Taylor (Royal Holloway and Bedford New College; Univ. of London; U.K) Bitnet/NetNorth/Earn: chaa006@vaxa.rhbnc.ac.uk (or) chaa006%rhbnc.vaxa@ac.uk (or) : chaa006@vaxb.rhbnc.ac.uk (or) chaa006%rhbnc.vaxb@ac.uk Arpa : chaa006%vaxa.rhbnc.ac.uk@ucl-cs.arpa (or) : chaa006%vaxb.rhbnc.ac.uk@ucl-cs.arpa