home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!darwin.sura.net!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!unixhub!tethys.SLAC.Stanford.EDU!ian
- From: ian@tethys.SLAC.Stanford.EDU (Ian A. MacGregor)
- Newsgroups: comp.databases.oracle
- Subject: Re: Hash clustering vs Indexing in ORACLE7
- Message-ID: <5399@unixhub.SLAC.Stanford.EDU>
- Date: 1 Sep 92 15:15:19 GMT
- References: <1992Aug27.115753.1300@hermes.dlogics.com> <z9hn27.tcox@netcom.com>
- Sender: news@unixhub.SLAC.Stanford.EDU
- Lines: 32
- Nntp-Posting-Host: tethys.slac.stanford.edu
-
- In article <z9hn27.tcox@netcom.com>, tcox@netcom.com (Thomas Cox) writes:
- |> mpd@hermes.dlogics.com writes:
- |> >Anybody used hash clustering in place of indexing?
- |> >Am considering it for a new app, and am wading through zillions of
- |> >pros and cons.
- |>
- |> >Thanks in advance...
- |>
- |> Why "in place of" -- I was under the impression you can do both. Though
- |> I have no idea what trade-offs _that_ entails...
- |>
- |> Cheers.
- |> -- Tom
- |> --
- |> Tom Cox DoD #1776 '91 CB 750 Nighthawk tcox@netcom.netcom.com
- |> My opinions! Mine! Not those of Netcom, or Oracle, or anybody else I know...
-
- I believe hashing is only available for clusters and not for individual tables.
- Hashing is faster for queries with predicates that do not include range scans.
- Range scans are more efficiently handled by B-Trees. Another problem is that
- hash functions do not guarnatee uniqueness; thus they are unsuitable for use as
- peimary keys in a table. Oracle does allow you to have both a hash and an
- index on your cluster key. It will use the hash for "=" queries and the
- B-Tree for range scans. There have been rumors about Oracle allowing hashing
- of individual table keys in the future, but I have heard no official statement
- from them.
-
-
- Ian MacGregor
- Stanford Linear Accelerator Center
- IAN@SLAC.STANFORD.EDU
- (415) 926-3528
-