LMS 3.0

Unanswered Question
Nov 26th, 2008

I have LMS 3.0 installed over windows 2003 server to mange 5000 node , I discovery 1300 node and can mange them and I have 4 gig ram but I have some problem

1- Very slow performance

2- I need to see the status for up and down node because DFM not get the normal device and topology service the normal behavior for poling is 2 hours for all node and the maximum in critical is 5 M for 30 node


What application in cisco works LMS 3.0 can get for me the up & down nods Real time monitor

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Joe Clarke Thu, 11/27/2008 - 09:34

Assuming you have all of LMS installed on this server, you would need Windows 2003 Enterprise Edition with 8 GB of RAM with PAE enabled, and four CPUs at 2.6 GHz or faster.

With just 1300 nodes, you would need 4 GB of RAM with a dual CPU box.

If performance is suffering, LMS will benefit from more RAM first, and more CPU second.

DFM is the application which provides fault management. It's ability to detect nodes going down in real-time is contingent on the following interval you configure. By default, polling occurs every four minutes. You can reduce this under DFM > Configuration > Polling and Thresholds > Polling Parameters. The lower you make the polling, the more CPU cycles DFM will use. This will require you to increase the CPU power and number of your machine to keep performance reasonable.

ahmed_saleh Fri, 11/28/2008 - 05:38


If I need to see the up & down devices .in DFM just I see the fault devices

In old CISCOWORKS LMS have the application can get the UP &DWON in Cisco works LMS 3.0 have this application or not if not what application can used in Cisco works to get for me all status for node UP& down

pvanvuuren Fri, 11/28/2008 - 05:29

I would consider splitting the LMS server modules over two servers with Common Services (DCR) in a Master and Slave setup. RME and Campus Manager on one and DFM and IPM on another.

Performance will improve greatly overall. With DFM, you might consider switching off the polling for interface utilization and then increase the polling for Availability. Start with 120 seconds and keep a watch on the performance of the server. Especially CPU. You want to make sure the poll cycle completes for all 5000 nodes before it starts a new one.

Take in consideration that DFM simply polls the devices (hopefully the Loopback Address) so if a Core node goes down and subsequent nodes attached to it (no redundancy), you will get an alert for each of them, thus if you have a need for event correlation you will need something else.


This Discussion