Ureader.com  
Microsoft software help and Community
   home   |   control panel login   |   archive   |  
 
Exchange
2000.active.directory
2000.admin
2000.announcements
2000.app.conversion
2000.applications
2000.clients
2000.clustering
2000.connectivity
2000.development
2000.documentation
2000.general
2000.information.store
2000.interop
2000.kms
2000.misc
2000.protocols
2000.realtime.collabo.
2000.setup
2000.transport
2000.win2000
admin
application.conversion
applications
clients
clustering
connectivity
design
development
misc
mobility
setup
tools
  
 
date: 28 Feb 2007 07:39:11 -0800,    group: microsoft.public.exchange.design        back       


Rate of change in an Exchange database   
We recently met with some consultants to discuss up and coming
Exchange design work.  I was very interested to hear that within a
database (in a normal working day) 80% of that file will change in a
day, even if there is little actual mail/change caused by user
activity.

Can anyone clarify this or point me maybe to an official document?
Many thanks, H.
date: 28 Feb 2007 07:39:11 -0800   author:   unknown

Re: Rate of change in an Exchange database   
On 28 Feb 2007 07:39:11 -0800, h_henry@yahoo.com wrote:

>We recently met with some consultants to discuss up and coming
>Exchange design work.  I was very interested to hear that within a
>database (in a normal working day) 80% of that file will change in a
>day, even if there is little actual mail/change caused by user
>activity.
>
>Can anyone clarify this or point me maybe to an official document?
>Many thanks, H.

That doesn't sound right.
Are you talking 80% of the file or 80% of the underlying blocks?
When you talk about rate of change you are typically talking about a
SAN based solution.
If you aren't running on a SAN then you aren't interested in the file
change since you either backup the entire store (full backup) and then
the logs (daily incrementals).
If you are talking about a SAN then the file has got nothing to do
with anything, bizzare as that might sound at first pass.

Do you want to elaborate a little?
date: Wed, 28 Feb 2007 16:01:08 +0000   author:   Mark Arnold [MVP]

Re: Rate of change in an Exchange database   
Unless you're doing something extremely unusual, like delivery to PST or all 
POP clients, that figure sounds very suspect.   I typically see more like 
3-5% at the item level.  It's important to distinguish that from the block 
level however.  I have seen aggressive Online Maintenance (online defrag 
represents the majority of changes) schedules cause block level changes on 
the order of 40%.  Do be aware that MS recommends that Online Defrag 
complete once every two weeks.  If you schedule less agressively to meet 
this target, then the block level changes are on the order of the item level 
changes.

John


 wrote in message 
news:1172677151.480155.33320@8g2000cwh.googlegroups.com...
> We recently met with some consultants to discuss up and coming
> Exchange design work.  I was very interested to hear that within a
> database (in a normal working day) 80% of that file will change in a
> day, even if there is little actual mail/change caused by user
> activity.
>
> Can anyone clarify this or point me maybe to an official document?
> Many thanks, H.
>
date: Wed, 28 Feb 2007 13:29:49 -0800   author:   John Fullbright [MVP] fjohn@donotspamnetappdotcom

Re: Rate of change in an Exchange database   
On Feb 28, 4:01 pm, "Mark Arnold [MVP]"  wrote:
> On 28 Feb 2007 07:39:11 -0800, h_he...@yahoo.com wrote:
>
> >We recently met with some consultants to discuss up and coming
> >Exchangedesign work.  I was very interested to hear that within a
> >database (in a normal working day) 80% of that file willchangein a
> >day, even if there is little actual mail/changecaused by user
> >activity.
>
> >Can anyone clarify this or point me maybe to an official document?
> >Many thanks, H.
>
> That doesn't sound right.
> Are you talking 80% of the file or 80% of the underlying blocks?
> When you talk aboutrateofchangeyou are typically talking about a
> SAN based solution.
> If you aren't running on a SAN then you aren't interested in the filechangesince you either backup the entire store (full backup) and then
> the logs (daily incrementals).
> If you are talking about a SAN then the file has got nothing to do
> with anything, bizzare as that might sound at first pass.
>
> Do you want to elaborate a little?

Interesting huh?

Thanks for your reply btw.

Yes - we are looking at a SAN based solution and we are looking at
having replicated copies/snaps versus tools that provide constant/
incremental mirrors like Replistor.  The "advantage" with a replistor
type of solution is that we would not have to ship full copies of the
databases every night, just changes.  I realise that these replicaiton
tools can have issues sometime and may need to "re-mirror" but we are
simply interested in our options.

The SAN consultants referred to this 80% rate of change figure, which
threw me in the meeting but I was really interested (and felt a little
silly for not "knowing").  Maybe as John Fullbirght has replied it is
to do with the online defrag.  The theory is that if 80% of change is
happening then an incremental tool will need to send 80% of data and
therefore you may as well use a SAN based copy tool.

Any thoughts would be appreciated.  My initial intention was to find
out if this was a known figure and if so where this is referenced...
Thanks.
date: 1 Mar 2007 01:51:31 -0800   author:   unknown

Re: Rate of change in an Exchange database   
On Feb 28, 9:29 pm, "John Fullbright [MVP]"
<fjohn@donotspamnetappdotcom> wrote:
> Unless you're doing something extremely unusual, like delivery to PST or all
> POP clients, that figure sounds very suspect.   I typically see more like
> 3-5% at the item level.  It's important to distinguish that from the block
> level however.  I have seen aggressive Online Maintenance (online defrag
> represents the majority of changes) schedules cause block level changes on
> the order of 40%.  Do be aware that MS recommends that Online Defrag
> complete once every two weeks.  If you schedule less agressively to meet
> this target, then the block level changes are on the order of the item level
> changes.
>
> John
>
>  wrote in message
>
> news:1172677151.480155.33320@8g2000cwh.googlegroups.com...
>
>
>
> > We recently met with some consultants to discuss up and coming
> >Exchangedesign work.  I was very interested to hear that within a
> > database (in a normal working day) 80% of that file willchangein a
> > day, even if there is little actual mail/changecaused by user
> > activity.
>
> > Can anyone clarify this or point me maybe to an official document?
> > Many thanks, H.- Hide quoted text -
>
> - Show quoted text -

Hi there, thanks for your reply.  We are not doing anything unusual
with our databases, but as my other reply states I am very interested
in getting the source from this quote.  Hence firstly I needed to know
if this was a known figure within the Exchange community (I'd not
heard it before, and so sure I would have) before going back to the
consultancy.  As we are looking at database copies across the SAN,
then we are talking block level I guess.  We have an online defrag
once a week, during the weekend, we don't have many users (3k) split
across two Exchange boxes.  Happy to hear any thoughts.  Thanks,
Hutton.
date: 1 Mar 2007 01:55:56 -0800   author:   unknown

Re: Rate of change in an Exchange database   
1.  If there are no changes, then online defrag has nothing to move.
2.  The first 10 steps of online maintenance (purging indices, marking 
items, etc) creates a great deal of IO, but impacts relatively few blocks. 
Snapping and shipping the change delta is very efficient for this type of 
IO.
3.  From Jeremey Kelley's Blog 
http://blogs.msdn.com/jeremyk/archive/2004/06/12/154283.aspx "The intention 
for online defragmentation is to free up pages in the database by compacting 
records onto the fewest number of pages possible, thus reducing the amount 
of I/O necessary. The ESE database engine does this by walking the database 
metadata (information in the database that describes tables in the database) 
and for each table, visit each page in the table and attempt to move records 
onto logically order pages. If you have an understanding of BTrees, we start 
at the furthest page to the right and begin compressing records to the left 
most page. This does not necessarily mean the pages are in order, but the 
movement is from a logical perspective."  If you make a lot of changes, and 
complete a defrag several times a day, you could see high rates of block 
level change.  The source for the 40% figure for online defrag was 
http://www.microsoft.com/technet/technetmag/issues/2006/10/FailoverClusters/default.aspx

If reduce the aggressiveness of online defrag, you quickly drop from 40% or 
so to 5 or 6% or lower.


 wrote in message 
news:1172742956.795866.177980@s48g2000cws.googlegroups.com...
> On Feb 28, 9:29 pm, "John Fullbright [MVP]"
> <fjohn@donotspamnetappdotcom> wrote:
>> Unless you're doing something extremely unusual, like delivery to PST or 
>> all
>> POP clients, that figure sounds very suspect.   I typically see more like
>> 3-5% at the item level.  It's important to distinguish that from the 
>> block
>> level however.  I have seen aggressive Online Maintenance (online defrag
>> represents the majority of changes) schedules cause block level changes 
>> on
>> the order of 40%.  Do be aware that MS recommends that Online Defrag
>> complete once every two weeks.  If you schedule less agressively to meet
>> this target, then the block level changes are on the order of the item 
>> level
>> changes.
>>
>> John
>>
>>  wrote in message
>>
>> news:1172677151.480155.33320@8g2000cwh.googlegroups.com...
>>
>>
>>
>> > We recently met with some consultants to discuss up and coming
>> >Exchangedesign work.  I was very interested to hear that within a
>> > database (in a normal working day) 80% of that file willchangein a
>> > day, even if there is little actual mail/changecaused by user
>> > activity.
>>
>> > Can anyone clarify this or point me maybe to an official document?
>> > Many thanks, H.- Hide quoted text -
>>
>> - Show quoted text -
>
> Hi there, thanks for your reply.  We are not doing anything unusual
> with our databases, but as my other reply states I am very interested
> in getting the source from this quote.  Hence firstly I needed to know
> if this was a known figure within the Exchange community (I'd not
> heard it before, and so sure I would have) before going back to the
> consultancy.  As we are looking at database copies across the SAN,
> then we are talking block level I guess.  We have an online defrag
> once a week, during the weekend, we don't have many users (3k) split
> across two Exchange boxes.  Happy to hear any thoughts.  Thanks,
> Hutton.
>
date: Thu, 1 Mar 2007 07:10:55 -0800   author:   John Fullbright [MVP] fjohn@donotspamnetappdotcom

Google
 
Web ureader.com


    COPYRIGHT 2007, YARDI TECHNOLOGY LIMITED, ALL RIGHT RESERVE  |   contact us