home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!emory!not-for-mail
- From: anasaz!qip.naomi@enuucp.eas.asu.edu (Naomi Walker)
- Newsgroups: comp.databases.informix
- Subject: Fragmentation
- Date: 21 Nov 1992 02:48:14 -0500
- Organization: Mailing List Gateway
- Lines: 60
- Sender: walt@mathcs.emory.edu
- Distribution: world
- Message-ID: <1ekpjuINNefh@emory.mathcs.emory.edu>
- Reply-To: anasaz!qip.naomi@enuucp.eas.asu.edu (Naomi Walker)
- NNTP-Posting-Host: emory.mathcs.emory.edu
- X-Informix-List-ID: <list.1632>
-
- > >
- > > In the good old days of SE, a table with a high level of activity
- > > (i.e. deletions/inserts) became riddled with old data row slots
- > > which were available for re-write, and the index began to look like
- > > swiss cheese. To correct this periodic unload/loads or cluster
- > > manipulation or ALTER TABLE (same as an unload/load but more user-friendly)
- > > were needed to keep index and data table efficiency at peak.
- > >
- > > Is this also true of Online? Or does Online do this automagically?
- > >
- > Raw Online space does get very fragmented, especially if you are short
- > of room in the partition. Table extents become very small and numerous
- > and it causes performance problems. Tables can generally have about
- > 240 extents, and it is possible to reach that value and come to a grinding
- > halt.
- >
- > We regularly monitor all our production systems extents and partition
- > fragmention. Tbunloading and tbloading the database back typically puts
- > all the data for each table in the first extent (if there is enough
- > contiguous room). We then carefully set all the next extents on the
- > tables, to accomodate future growth. If tbunloading is unreasonable
- > because of the database size, we will create an empty new table, with
- > the first extent set, and copy the data from the old table to the
- > new table.
- >
- >
- >
- Thank you, I got a slightly different answer yesterday to wit:
-
- ->Online does automatic page compression when it writes disk. the tbstat -p
- ->command will show the number of compresses that have been executed. see
- ->page 2-117 in the dba manual (v5.0).
- ->
- ->regards,
-
- It sounds like you are discussing disk fragmentation which is another HIGHLY
- viable concern and I thank you for your note.
-
- Your welcome. I try to pass on what I have learned from actual experience
- with the beast. We are currently running Online 4.1 on our production
- systems. I wonder if 4.1 employs the same technique described in the 5.0
- manual. I know there are many features 4.1 supports that were not documented
- until 5.0. I have the 5.0 DBA manual for a preview and it *is* an outstanding
- manual.
-
- Upon reading the chapter on data compression, it appears to me that they
- are describing compression within a page, and that information is good
- to understand. I was speaking in a larger sense about the entire partition.
-
- Have fun,
-
- Naomi
-
-
-
- --
- Naomi Walker (aka N7FSA) naomi%anasaz.UUCP@asuvax.eas.asu.edu
-
- Enthusiasm is caught, In tennis,
- not taught. love means nothing.
-