It seems that Netapp is very confused about
this issue and or topic, as I have called into tech support on three separate occasions
about this issue and I'm always left feeling more confused than when I called
in. Netapp needs to figure this out and give an answer to this question, which I
think should be a YES or NO answer. Can the waffle scan safely defrag an exchange
DB, if so what are the steps? If a waffle scan cannot defrag an exchange DB then
can running eseutil and pointing eseutil to the exchange DB on the filler run,
and will it defrag the DB?
Thanks
Skip
From: Waters, G Scott
THE LOG.SEC TEAM [mailto:george.scott.waters@us.army.mil]
Sent: Monday, June 07, 2004 7:33
AM
To:
Subject: RE: Defraging the
excahgne DB
Importance: High
WOW ...
this is the exact opposite of what NetApp support told me.
On a
volume used for storing Exchange data, the WAFL SCAN REALLOCATE command will
never complete. It goes for a while and then stalls.
We
monitor the layout ratios weekly and schedule down time to perform an offline
DEFRAG every 4 - 6 weeks.
This is
all based on information received at the time ... If NetApp has fixed this then
someone enlighten me.
- Scott
W.
-----Original
Message-----
From:
Sent: Thursday, June 03, 2004 2:31
PM
To: 'Borders, Rich';
toasters@mathworks.com
Subject: RE: Defraging the excahgne
DB
Hi
I just
got done talking with support at Netapp, and they told me that doing
an offline defrag of my exchange DB
using the esutil will have absolutely no
impact on the exchange DB on the
filer, and they also told me that in order
to defrag the exchange DB I must
use the fillers Waffle utility to do this
and that this would also defragment
everything on the vol. They said that
it's the responsibility of the file
system on the filer to determine if a
vol is fragmented or not, this
includes all data on the vol. So knowing this
is it safe to defrag a vol that
contains an exchange DB??
Thanks
again for your help it is much appreciated
Skip
-----Original
Message-----
From: Borders, Rich [mailto:Rich.Borders@netapp.com]
Sent: Thursday, June 03, 2004 11:13
AM
To:
Subject: RE: Defraging the excahgne
DB
<<<
I am replying in-line >>>
From:
Sent: Thursday, June 03, 2004 1:33 PM
To: toasters@mathworks.com
Subject: Defraging the excahgne DB
Howdy
I have my exchange
5.5 DB pub and priv and my filler it's a
FAS250 and I want to know a few
things
#1 can I run the
eseutil utility if the DB is on my FAS 250 to
defrag the DB?
<<<
This is doable as it is an offline defrag. The curious thing about
doing this is that you are
accomplishing the APPLICATION defrag, and at
the same time getting a great
by-product of a WAFL defrag. ( Essentially
a database re-write ) >>>
#2 is it the filer
that determines if the exchange DB is
fragmented or not, and if it is how
would I defrag the DB?
<<<The
Application can fragment ( very common and accounted for by an
online and offline defrag funciton
in Exchange ) and the Filer can
fragment. I say that tounge in
cheek because the fragmentation is really
only identified when you access the
data. Since the database is one
large file the request for data in
it is atypical, the data 'fragments'
over time. The blocks of the filer
rarely match up to the blocks of the
database, so the issues becomes
when you want to get a LOT of sequential
data from a large file ( aka mail
store verify ) you will be essentially
asking the filer to scan through
it's blocks at a super high rate of
speed.>>>
#3 what happens to
all other data that is stored on the same vol
as the exchange DB once I defrag
this vol? ie snap shots
<<<The
Mail store can be reallocated or the volume can. I would
recommend the snapshots be removed
as you can really get yourself into a
pickle by 'RE'-writing a database
with an application defrag, or a WAFL
reallocate. If you are short on
space, you need to get the snapshots out
of there. Snapshots are locked
images and if the file system is
'fragmented' then it is unlikely that
you are doing much more than
spinning your wheels if you are
trying to defragment into a fragemnted
snapshot. >>>
#4 I ran a waffle
scan and I got a reading of 1.8 si this
fragmented?
That is
basically a ratio of actual file size ( df essentially ) to
block allocation. Ex. A one
terabyte volume has a 500 gig database. If
it's layout is 1 then the File is
essentially using the exact number of
blocks that a Filer of 500 gigs
needs. If the layout were 1.8 then the
500 gig df would still be 500 but
the block allocation would actually be
closer to the 900 gig . This is
usually not an issue until the ratio
takes the block allocation higher
then the volume size. A 2.1 layout
would mean that a 500 gig df would
be 1.05 TB. This means you are
rescanning on a request of data on
the last .05 TB. If the Filer were a
900 gig df, then a layout of 1.8
would be MUCH more impactful. It is not
the end of the world, you just have
to manage it. You are on the right
track I can see, so good luck.
Thanks a bunch
Skip
Rich
Borders