There isn't any enforced limit to the NotifyBuffer setting but for
practical purposes it shouldn't be bigger
than 4 meg.

The file system monitor portion of the journal service records real time
file system change notifications
into a preallocated buffer (a requirement of the Microsoft API used to
monitor the file system) so there is
a finite limit to the volume of activity the monitor is capable of
processing.

Basically if the notifications arrive faster than the monitor can process
them the notification buffer will
overflow and the journal will be invalidated and a normal full incremental
will have to be performed
before Journal Based Backup can continue.

That having been said, our (TSM development and System Test) testing has
shown that most change
activity generated in a typical production environment can be journaled
with a reasonable size notification
buffer (4 meg or less).

We have written test programs  generating  very large amounts of change
activity in very
short windows in time which consistently overflow the notification buffer,
but we have come to
the conclusion that these programs do not really simulate the change
activity of a
typical production environment.

Copying or deleting very large directory trees may also generate enough
change activity
to overflow the buffer.

Journal Based Backup is most beneficial in environments with very large
file systems which
generate a small to moderate (5 to 25 percent) amount of change activity on
a regular basis.

Environments which generate very large amounts of change activity
(especially mass
copy and delete operations) in very short periods of time are probably not
going to
benefit from Journal Based Backup.

Hope this helps answer your questions and please post any further ones you
may have.


Pete Tanenhaus
Tivoli Storage Solutions Software Development
email: tanenhau AT us.ibm DOT com
tieline: 320.8778, external: 607.754.4213

"Those who refuse to challenge authority are condemned to conform to it"

---------------------- Forwarded by Pete Tanenhaus/San Jose/IBM on
01/04/2002 04:39 PM ---------------------------
01/04/2002 04:39 PM ---------------------------

"Scott, Brian" <bscott AT EDS DOT COM>@VM.MARIST.EDU> on 01/04/2002 11:15:57 AM

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

Sent by:    "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>


To:    ADSM-L AT VM.MARIST DOT EDU
cc:
Subject:    TSM 4.2.1 Journal Settings



Good Mornin' *SMer's,

Does anyone know what's the max setting for the NotifyBufferSize= in the
tsmjbbd.ini? We're running the Journal Service on a TSM 4.2.1 Win2k box
with
3.5GB of RAM. We have 2 filesystems besides the O/S that are about 280GB
apiece comprised of 2.3 million objects apiece. I tried a touch command on
25% of the filesystem to simulate change to the files to populate the
journal but received the following error during the file change process:

01/04/2002 10:07:00 psFsMonitorThread(tid 2688): Notification buffer
overrun
for monitored FS 'R:\'.

I tried changing the NotifyBufferSize= 0x3200000 but didn't work.

Anyone have any ideas or know the max setting?

Thanks,

Brian Scott
EDS - Enterprise Distributed Capabilities
MS 3278
Troy, MI 48098

* phone: 248-265-4596 (8-365)
* mailto:bscott AT eds DOT com