Veritas-bu

[Veritas-bu] Schedules Question.

2002-05-29 13:30:40
Subject: [Veritas-bu] Schedules Question.
From: Mark.Donaldson AT experianems DOT com (Donaldson, Mark)
Date: Wed, 29 May 2002 11:30:40 -0600
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_01C20736.8DAC4160
Content-Type: text/plain

Cummulative incrementals can save you from tape failure more than a
differential can. 

If you take a full backup once a month, then do differentials nightly from
there, you risk not being able to restore to current date if you lose an
intermediate differential.  

Look at this example

Date  Type 
01    Full
02    Diff
03    Diff
...
20    Diff
21    Diff

Now assume you lose a disk set on the 22nd and you want to restore to the
last date.  You've got to put in a Full backup from the 1st of the month,
then apply 21 differentials.  This can take a lot of time.

Now assume that due to tape failure you can't restore the differential from
the 10th.  You can only restore the system accurately to the 9th.  All the
other differentials from the 9th are based on changes from the 10th and
that's not restorable.  The failure of an intermediate tape invalidates all
subsequent differentials. You can't say the the user that the system is up
to date - you can say that it's all back except that any files changed on
the 10th are gone - if I were a customer, I'd think toss that statement on
my "This is a bad thing" pile.

If you are doing cummulative differentials each night, you'll be backing up
more data.  However, to restore to any date, you need the Full backup, and
the last cummulative.  You'll restore two images, not twenty-something so
it's faster.  Also, any failure of an intermediate cummulative image has no
affect on following images as each cummulative image stands by itself.  Your
mixed diff/cumm method also allows a Full restore followed by a Cummulative,
followed up by no more than 6 differentials. Still pretty good compared to
20-something possible images using only nightly differentials.

This being said, since NB will colocate identical retention images, you have
to think ahead - a cummulate backup technically stands by itself but only if
a lost tape doesn't contain the previous night's images.  A botched tape
will lose multiple nights if multiple images are on it.  My most critical
backups go to a two different pools on alternating nights.  This way if I
lose a tape, the previous night's images are not on that same tape.   The
exposure to colocating images on tape with a differential backup is even
bigger; alternating tape pools will not help you on this.

So, to sum up.
Differential - less tape therefore cheaper - more exposure to media failure
- longer restore times.
Cummulative - more tape therefore more expensive - more protection from
media failure - shorter restore times.

Your current method looks pretty good at protecting your customer's data,
since you have a cummulative protecting against a differential failure.

HTH - Mark

-----Original Message-----
From: Christopher Manders [mailto:CJManders AT lbl DOT gov]
Sent: Tuesday, May 28, 2002 5:16 PM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Schedules Question.


Hi,

So, we have been looking at how we were implemented, as we are starting to
use
_alot_ of tapes.

Basically, what we have is:
Daily-Differentials
Weekly-Cumulative Incrementals
Monthly -Fulls

Is there a real advantage to having weekly cumulatives, as our customers,
who
we have to charge, complain about the extra tape usage?

Basically, we could really use the saved tapes. But, I wanted other people's
experiences before I decided.

Thanks for any input or suggestions.


Chris

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

------_=_NextPart_001_01C20736.8DAC4160
Content-Type: text/html
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=3DUS-ASCII">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: [Veritas-bu] Schedules Question.</TITLE>
</HEAD>
<BODY>

<P><FONT SIZE=3D2>Cummulative incrementals can save you from tape =
failure more than a differential can. </FONT>
</P>

<P><FONT SIZE=3D2>If you take a full backup once a month, then do =
differentials nightly from there, you risk not being able to restore to =
current date if you lose an intermediate differential.&nbsp; =
</FONT></P>

<P><FONT SIZE=3D2>Look at this example</FONT>
</P>

<P><FONT SIZE=3D2>Date&nbsp; Type </FONT>
<BR><FONT SIZE=3D2>01&nbsp;&nbsp;&nbsp; Full</FONT>
<BR><FONT SIZE=3D2>02&nbsp;&nbsp;&nbsp; Diff</FONT>
<BR><FONT SIZE=3D2>03&nbsp;&nbsp;&nbsp; Diff</FONT>
<BR><FONT SIZE=3D2>...</FONT>
<BR><FONT SIZE=3D2>20&nbsp;&nbsp;&nbsp; Diff</FONT>
<BR><FONT SIZE=3D2>21&nbsp;&nbsp;&nbsp; Diff</FONT>
</P>

<P><FONT SIZE=3D2>Now assume you lose a disk set on the 22nd and you =
want to restore to the last date.&nbsp; You've got to put in a Full =
backup from the 1st of the month, then apply 21 differentials.&nbsp; =
This can take a lot of time.</FONT></P>

<P><FONT SIZE=3D2>Now assume that due to tape failure you can't restore =
the differential from the 10th.&nbsp; You can only restore the system =
accurately to the 9th.&nbsp; All the other differentials from the 9th =
are based on changes from the 10th and that's not restorable.&nbsp; The =
failure of an intermediate tape invalidates all subsequent =
differentials. You can't say the the user that the system is up to date =
- you can say that it's all back except that any files changed on the =
10th are gone - if I were a customer, I'd think toss that statement on =
my &quot;This is a bad thing&quot; pile.</FONT></P>

<P><FONT SIZE=3D2>If you are doing cummulative differentials each =
night, you'll be backing up more data.&nbsp; However, to restore to any =
date, you need the Full backup, and the last cummulative.&nbsp; You'll =
restore two images, not twenty-something so it's faster.&nbsp; Also, =
any failure of an intermediate cummulative image has no affect on =
following images as each cummulative image stands by itself.&nbsp; Your =
mixed diff/cumm method also allows a Full restore followed by a =
Cummulative, followed up by no more than 6 differentials. Still pretty =
good compared to 20-something possible images using only nightly =
differentials.</FONT></P>

<P><FONT SIZE=3D2>This being said, since NB will colocate identical =
retention images, you have to think ahead - a cummulate backup =
technically stands by itself but only if a lost tape doesn't contain =
the previous night's images.&nbsp; A botched tape will lose multiple =
nights if multiple images are on it.&nbsp; My most critical backups go =
to a two different pools on alternating nights.&nbsp; This way if I =
lose a tape, the previous night's images are not on that same =
tape.&nbsp;&nbsp; The exposure to colocating images on tape with a =
differential backup is even bigger; alternating tape pools will not =
help you on this.</FONT></P>

<P><FONT SIZE=3D2>So, to sum up.</FONT>
<BR><FONT SIZE=3D2>Differential - less tape therefore cheaper - more =
exposure to media failure - longer restore times.</FONT>
<BR><FONT SIZE=3D2>Cummulative - more tape therefore more expensive - =
more protection from media failure - shorter restore times.</FONT>
</P>

<P><FONT SIZE=3D2>Your current method looks pretty good at protecting =
your customer's data, since you have a cummulative protecting against a =
differential failure.</FONT></P>

<P><FONT SIZE=3D2>HTH - Mark</FONT>
</P>

<P><FONT SIZE=3D2>-----Original Message-----</FONT>
<BR><FONT SIZE=3D2>From: Christopher Manders [<A =
HREF=3D"mailto:CJManders AT lbl DOT gov">mailto:CJManders AT lbl DOT 
gov</A>]</FONT>
<BR><FONT SIZE=3D2>Sent: Tuesday, May 28, 2002 5:16 PM</FONT>
<BR><FONT SIZE=3D2>To: veritas-bu AT mailman.eng.auburn DOT edu</FONT>
<BR><FONT SIZE=3D2>Subject: [Veritas-bu] Schedules Question.</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>Hi,</FONT>
</P>

<P><FONT SIZE=3D2>So, we have been looking at how we were implemented, =
as we are starting to use</FONT>
<BR><FONT SIZE=3D2>_alot_ of tapes.</FONT>
</P>

<P><FONT SIZE=3D2>Basically, what we have is:</FONT>
<BR><FONT SIZE=3D2>Daily-Differentials</FONT>
<BR><FONT SIZE=3D2>Weekly-Cumulative Incrementals</FONT>
<BR><FONT SIZE=3D2>Monthly -Fulls</FONT>
</P>

<P><FONT SIZE=3D2>Is there a real advantage to having weekly =
cumulatives, as our customers, who</FONT>
<BR><FONT SIZE=3D2>we have to charge, complain about the extra tape =
usage?</FONT>
</P>

<P><FONT SIZE=3D2>Basically, we could really use the saved tapes. But, =
I wanted other people's</FONT>
<BR><FONT SIZE=3D2>experiences before I decided.</FONT>
</P>

<P><FONT SIZE=3D2>Thanks for any input or suggestions.</FONT>
</P>
<BR>

<P><FONT SIZE=3D2>Chris</FONT>
</P>

<P><FONT =
SIZE=3D2>_______________________________________________</FONT>
<BR><FONT SIZE=3D2>Veritas-bu maillist&nbsp; -&nbsp; =
Veritas-bu AT mailman.eng.auburn DOT edu</FONT>
<BR><FONT SIZE=3D2><A =
HREF=3D"http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu"; =
TARGET=3D"_blank">http://mailman.eng.auburn.edu/mailman/listinfo/veritas=
-bu</A></FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C20736.8DAC4160--

<Prev in Thread] Current Thread [Next in Thread>