|
[Veritas-bu] Compressibility of Oracle Databases
2002-08-09 18:47:59
This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.
------_=_NextPart_001_01C23FF6.CF606160
Content-Type: text/plain;
charset="iso-8859-1"
Just a quick note to broadcast these results since the question has come up.
I compared the compressibility of Oracle data files by turning on the
Netbackup client compression and running two backups of two different
databases. This was compared against the uncompressed backup runs on the
following evening. I then compared the compression ratio of the images sets
against the ratio of allocated to occupied data blocks.
The raw data looks like this:
DB1 - 16% occupied blocks, overall compression ratio 36%
DB2 - 71% occupied blocks, overall compression ratio 77%
Assuming that the compressibility of data blocks is consistent between the
databases, I set up two equations:
16D + 84Z = 36
71D + 29Z = 77
...where D & Z are Data block & Zero block compression rates respectively.
solving using some high-school algebra I get the following:
Data block compression 0.99 to 1 - doesn't squish down worth nothin'
(actually 98.6%)
Empty block compression: 4:1 - pretty good (actually 24.1%)
This may allow some quick-and-dirty estimations for various databases. For
example, if you've got a 100G allocated database that's 75% occupied, then:
75G * .99 + 25G * 0.24 = 79.3 Gig of tape occupied.
Caveats: this was for NB client compression, not hardware compression at the
tape drive. I'm assuming that the algorithms are close enough to generalize
to overall compressibility.
-Mark
-----------------------------------------------------------------------
Mark Donaldson - Sr. Systems Engineer
Experian - Denver Colorado
-----------------------------------------------------------------------
There are 10 kinds of people in this world, those
that understand binary and those that don't.
-----------------------------------------------------------------------
------_=_NextPart_001_01C23FF6.CF606160
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>Compressibility of Oracle Databases</TITLE>
</HEAD>
<BODY>
<P><FONT SIZE=3D2>Just a quick note to broadcast these results since =
the question has come up.</FONT>
</P>
<P><FONT SIZE=3D2>I compared the compressibility of Oracle data files =
by turning on the Netbackup client compression and running two backups =
of two different databases. This was compared against the =
uncompressed backup runs on the following evening. I then =
compared the compression ratio of the images sets against the ratio of =
allocated to occupied data blocks.</FONT></P>
<P><FONT SIZE=3D2>The raw data looks like this:</FONT>
<BR><FONT SIZE=3D2>DB1 - 16% occupied blocks, overall compression ratio =
36%</FONT>
<BR><FONT SIZE=3D2>DB2 - 71% occupied blocks, overall compression ratio =
77%</FONT>
</P>
<P><FONT SIZE=3D2>Assuming that the compressibility of data blocks is =
consistent between the databases, I set up two equations:</FONT>
</P>
<P><FONT SIZE=3D2>16D + 84Z =3D 36</FONT>
<BR><FONT SIZE=3D2>71D + 29Z =3D 77</FONT>
</P>
<P><FONT SIZE=3D2>...where D & Z are Data block & Zero block =
compression rates respectively.</FONT>
</P>
<P><FONT SIZE=3D2>solving using some high-school algebra I get the =
following:</FONT>
</P>
<P><FONT SIZE=3D2>Data block compression 0.99 to 1 - =
doesn't squish down worth nothin' (actually 98.6%)</FONT>
<BR><FONT SIZE=3D2>Empty block compression: 4:1 - pretty good (actually =
24.1%)</FONT>
</P>
<P><FONT SIZE=3D2>This may allow some quick-and-dirty estimations for =
various databases. For example, if you've got a 100G allocated database =
that's 75% occupied, then:</FONT></P>
<P><FONT SIZE=3D2>75G * .99 + 25G * 0.24 =3D 79.3 Gig of tape =
occupied.</FONT>
</P>
<P><FONT SIZE=3D2>Caveats: this was for NB client compression, not =
hardware compression at the tape drive. I'm assuming that the =
algorithms are close enough to generalize to overall =
compressibility.</FONT></P>
<P><FONT SIZE=3D2>-Mark</FONT>
</P>
<P><FONT =
SIZE=3D2>---------------------------------------------------------------=
--------</FONT>
<BR><FONT SIZE=3D2> Mark Donaldson - Sr. Systems =
Engineer</FONT>
<BR><FONT SIZE=3D2> Experian - Denver Colorado</FONT>
<BR><FONT =
SIZE=3D2>---------------------------------------------------------------=
-------- </FONT>
<BR><FONT SIZE=3D2> There are =
10 kinds of people in this world, those</FONT>
<BR><FONT =
SIZE=3D2> that =
understand binary and those that don't. </FONT>
<BR><FONT =
SIZE=3D2>---------------------------------------------------------------=
--------</FONT>
</P>
</BODY>
</HTML>
------_=_NextPart_001_01C23FF6.CF606160--
|
|
|
