I do not know if there are any architectural or product limitations. The
only limit that I have encountered is strictly a practical one. We are
running about 25 copies of the MLM product on Sun Sparc Ultra-2 servers with
dual 300MHz processors and 512 MB of memory. We are only using the
distributed status polling and a little data collection and thresholding.
We have had a problem on only one of the servers that is polling the most IP
addresses, about 1800. The problem is that the MLM application does not
respond to NetView's "heartbeat poll", which results in the MLM Down trap,
and then NetView dynamically takes over status polling. I suspect that the
problem is related to the number of IP addresses, because it works fine when
I reduce them to about 1500 IP addresses. I am working on upgrading to MLM
5.0.3 (currently on 5.0.1) and Solaris 2.7 (currently 2.5.1) in hopes that
the server/application will perform better. The problem is not related to
high CPU utilization or anything like that either. I suspect that the
midmand daemon is just so busy doing its normal work that it does not
respond to the NetView poll in a timely manner.
> -----Original Message-----
> From: Frank W. Hansen [SMTP:fhansen@ENTERPRISEINNOVATIONS.COM]
> Sent: Sunday, September 05, 1999 19:56
> To: NV-L@UCSBVM.UCSB.EDU
> Subject: MLM question
>
> What is the maximum number of devices that a MLM can support?
>
> Best Regards,
>
> Frank W. Hansen
> Tivoli Enterprise Certified Consultant
>
> Enterprise Innovations
> www.enterpriseinnovations.com <http://www.enterpriseinnovations.com>
> 910 458 9172
|