What's flawed is that the people in charge of NetView are not in charge of
the network. Or that the people in charge of the network are not
responsible for making it efficient. ICMP is an exceedingly efficient way
of determining status. SNMP takes much more bandwidth -- bigger packets,
more cpu by both agents, and so on -- which means that more of the network
capacity is given over to managing the network and the real users have less
throughput. We don't care if you want to run an inefficient network, but
you need to understand just what the cost is.
If ICMP packets are being lost then you can bet user packets are also being
lost and people are experiencing higher response times as a result. So
those "false downs" you get when the pings fail aren't really false. They
are telling you that there is a network bottleneck that needs to be
addressed by tuning. Sure you can eliminate the intermittent red colors on
your map by determining status for everything via SNMP. But you are
masking a different problem.
My two cents.
James Shanks
Team Leader, Level 3 Support
Tivoli NetView for UNIX and NT
danny@UK.IBM.COM@tkg.com on 03/21/2001 03:50:14 AM
Please respond to IBM NetView Discussion <nv-l@tkg.com>
Sent by: owner-nv-l@tkg.com
To: IBM NetView Discussion <nv-l@tkg.com>
cc:
Subject: [NV-L] Re: OIDs Seedfiles and SNMP status polling
James,
The reason why is quite simple - the managed network that my customer uses
loses pings very very often, and the network provider is not doing much to
address it. Our logic is that as SNMP is a higher priority than ICMP, there
is less chance of a false Node or Interface Down event.
Please tell me if our logic is flawed.
Cheers,
Danny
P.S. Thanks Sylvia for reminding me about the P flag in the oid_to_type
file - I saw it a few weeks ago and then promptly forgot it!
Date: Tue, 20 Mar 2001 09:19:33 -0500
From: James_Shanks@tivoli.com
Subject: Re: OIDs Seedfiles and SNMP status polling
Just a question on why you would want to do this.
Since the overhead involved to get status via SNMP rather than by ping is
so much greater, why would you want to make that the default? Getting
status via SNMP is a very new feature in NetView and it should be used
sparingly, which probably accounts more than anything else for why you
cannot simply set things up as you suggest. I think product management
would want to see a full-blown performance analysis before they'd let such
a feature out the door. But that is just my opinion here.
If you think this should be a feature, then my suggestion is that you call
Support and open an enhancement request.
James Shanks
Team Leader, Level 3 Support
Tivoli NetView for UNIX and NT
Danny Williams
Tivoli & AIX Specialist
IBM Global Services - Integrated Technology Services
mailto:danny@uk.ibm.com
Tel: 665520 (01926-465520) -- MOBX: 275520 (07967-275520)
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
|