Active IQ Unified Manager Discussions
Active IQ Unified Manager Discussions
We are working on upgrading our DFM infrastructure to v4.0.2. We want to ensure extremely fast response time for DFM and we would like to integrate into our existing reporting systems. Have some questions:
1. It appears DFM 4.0.2 is still 32-bit. Is there any advantage to giving the system more than 4GB of RAM if it is only a 32-bit database engine?
2. What is the maximum value you can set for the dfm dbCacheSize option that would help with performance? The manual recommends 1024, but if you have 16GB of RAM, is there any performance value in setting this higher.
3. We are running on RHEL 5.6 64-bit. Are there any kernel or cache settings that can help increase DFM performance?
If we have to scale out to ensure good DFM performance:
4. How can we connect to the DFM database to retrieve the information into our reporting systems? Is there an ODBC connector available? Is the data structure published?
5. Can we use snapshots to backup and replicate the database? What command can we run to quiesce the database for the snap?
Thanks in advance!
Ted
I may have some answers...
4. DB structure
A DB Admin told me once, that it is very easy to gain access to the DB structure and the data in there.
Maybe this part of the documentation helps?
5. Backup
In the DFM manual (http://now.netapp.com/NOW/knowledge/docs/DFM_win/rel402/html/software/opsmgr/GUID-893CC249-67DC-4F76-9345-DD834C8726BC.html😞
What the DataFabric Manager database backup process is
DataFabric Manager database backup is simple and powerful. You can back up the DataFabric
Manager database, script plug-ins, and performance data without stopping any DataFabric Manager
services. However, data collection and view modifications of Performance Advisor are suspended
during the backup process.
There are two types of backups:
• Archive: The backup process backs up your critical data in compressed form using the ZIP
format. The DataFabric Manager server automatically converts the DataFabric Manager data to
an archive format and stores the backup in a local or remote directory. You can easily move an
archive-based backup to a different system and restore it. But the backup process is timeconsuming.
• Snapshot-based: In the Snapshot-based approach, the backup process uses the Snapshot
technology to back up the database. You can quicken the backup process through this approach.
But you cannot easily move a Snapshot-based backup to a different system and restore it.
Peter
Hi, welcome to the community.
I am posting because I am keen to see what replies you get. We have played with the DB settings in the past but it always stopped the service from starting. The key performance bottleneck for us has always been disk I/O on the DFM server as we run it on local disk (DAS). I know Oncommand (DFM 2) is due out soon and a lot of performance work has been done, so maybe talk to your NetApp SE to see if this may be a better option going forward.
Hope it helps
Bren
Hi Ted,
Pls find my responses inline.
We are working on upgrading our DFM infrastructure to v4.0.2. We want to ensure extremely fast response time for DFM and we would like to integrate into our existing reporting systems. Have some questions:
1. It appears DFM 4.0.2 is still 32-bit. Is there any advantage to giving the system more than 4GB of RAM if it is only a 32-bit database engine?
It true that the DFM v4.0.2 is a 32 bit application, but it has 6 process, each of which consumes upto 1.8G,
[root@lnx ~]# dfm service list
sql: started
http: started
eventd: started
monitor: started
scheduler: started
server: started
watchdog: started
[root@lnx ~]#
Its generally the sql, server, monitor, and eventd tend to consume upto 1.8G each. So have at least 12 GB for dfm and another 4 G for operating system.
Also the next release of DFM called OnCommand 5.0 is a pure 64 bit application and it scales along with the provided hardware.(both in terms of RAM,Cores&CPU).
2. What is the maximum value you can set for the dfm dbCacheSize option that would help with performance? The manual recommends 1024, but if you have 16GB of RAM, is there any performance value in setting this higher.
When you have 16 GB you can leave it at default, as we take half of the 1.8G for dbCacheSize, this recommendation hold good when you have less RAM.Either set it at 1G or leave it at
default.
3. We are running on RHEL 5.6 64-bit. Are there any kernel or cache settings that can help increase DFM performance?
If you are planning to run Protection/Provisioning Manager, pls increase the number for semaphore from 128 to 1024.
On Red Hat Linux, the defaultlimit of 128 semaphore arrays can be increased to 1024 by adding line below to /etc/sysctl.conf:
kernel.sem=250 32000 32 1024
Where
250 max semaphores per array
32000 maxsemaphores system wide
32 max number of operations per semop call
1024 value for the number of semaphore arrays ---this what needs to be modified.
4. How can we connect to the DFM database to retrieve the information into our reporting systems? Is there an ODBC connector available? Is the data structure published?
Yes,Database schema is documented under the following link in your installation.
Open Operation Manager WebUI->ControlCenter->Help->General Help->Database Schema.
Below is the link to the doc on how to access the db views using third party tools.
http://media.netapp.com/documents/tr-3690.pdf
5. Can we use snapshots to backup and replicate the database? What command can we run to quiesce the database for the snap?
Yes,DFM has its support for DR, and all this can be done.
Below is the link for the same.
Regards
adai
Message was edited by: Adaikkappan Arumugam