home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.unix.internals:2103 comp.arch.storage:907 comp.unix.large:414 comp.unix.wizards:5399
- Path: sparky!uunet!ferkel.ucsb.edu!taco!lll-winken!ames!saimiri.primate.wisc.edu!zaphod.mps.ohio-state.edu!menudo.uh.edu!nuchat!lobster!siswat!buck
- From: buck@siswat.hou.tx.us (Lester Buck)
- Newsgroups: comp.unix.internals,comp.arch.storage,comp.unix.large,comp.unix.wizards
- Subject: large directory algorithms - B-tree?
- Message-ID: <1993Jan10.051038.6931@siswat.hou.tx.us>
- Date: 10 Jan 93 05:10:38 GMT
- Organization: Photon Graphics
- Lines: 13
-
- A colleague is designing a Unix name server (effectively namei) which
- needs to scale to very large (possibly to 10**12) numbers of names.
-
- I remember once seeing a passing reference to an implementation of Unix
- directory algorithms which used B-trees. It was described as a failure,
- as the complexity of a full blown B-tree was not necessary for typical
- Unix directory sizes. Can anyone supply a reference to this work, such
- as a Usenix paper?
-
- Thanks alot!
-
- --
- A. Lester Buck buck@siswat.hou.tx.us ...!uhnix1!siswat!buck
-