Hi Magnus,

When it comes to monitoring throughput on routers and switches, MRTG along
with RRD is the natural choice. First of all it is free, second there is a
tremendous community to help support. In its current incarnation, WUG does
not do throughput trending. It can however monitor certain thresholds on
interfaces (ifInOctet and ifOutOctet). Most likely you want to have a look
at the trends over time for a particular interface. In one of your earlier
posts you asked about Denika. Well, Denika does exactly that (and lots
more). It does use MRTG/RRD and gives you a fairly easy way of doing
trending. It does not stop at interfaces, it can trend any SNMP variable
over time. As a matter of fact, with a little help from a script, it can
trend anything over time. If you don't need the nice interface and want to
dig a bit deeper into trending, MRTG/RRD can do all of this under Windows
and under Linux. See http://people.ee.ethz.ch/~oetiker/webtools/mrtg/ for
more info.

What also could be done is to report on bandwidth usage. One would need to
count the bits and bytes over time. Again, this can be done with MRTG/RRD
and a little scripting work.

When it comes to identifying which traffic goes where, things are a bit more
involved. Cisco has something like sflow where when processed, one could
identify what kind of traffic went where and when. This is particular
interesting when you have several critical applications fighting for the
same bandwidth. Unfortunately, whilst sflow gains popularity, every
router/switch vendor does it slightly different.

I hope this gives you some more pointers.

Luz Berger
Berger Network Consult
http://www.bergerl.com


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Finbom Magnus
Sent: Friday, July 30, 2004 12:25 PM
To: '[EMAIL PROTECTED]'
Subject: SV: [WhatsUp Forum] How to know what to look for?


Thanks you very much Luz!

You are right on that it really not is up to the IT-departement but the
needs of the business that controlls. An SLA is a good start, even for a
small organization just to point out what's the more important. But in my
case the organization is very poor in telling a realistic goal. Of course I
too want everything to be up 24/7... 

Althogh I have been thinking of trying once again to ask. As it is now my
curiosity (and some free time) of whats can be done thats guides me on what
we should monitor. When I started at this work they had no monitoring at
all. And they were satisfied when I gave the WUG with the simple up/down
alerts.

My experience in computing goes way back but SNMP is something I hav ekept
away from becuse of lack of time. For a server it is easy for me to state
that most of the parts can be monitored with SNMP. I just asume that
everything that the system log in the eventlogg are also possible to get
through SNMP.

For a switch I think that port-states, errors can be monitored.. But im
unsure wheter its possible to monitor bandwith-load.. But some reading about
it will tell me. (Or asking here in this forum :-)

Im realizing that monitoring can take a full time job. And if monitoring
everything and creating reports more than lots of time would be needed. SNMP
is hard a first, but very "exciting" when getting into it. And just when
starting to figure out that It not was as difficult as I thought I
understand that it is very time-consuming and demanding of lots of R.T.F.M.

I will save and search through the links you provided and will hopefully
find many usefull things.


Regards, Magnus Finbom


-----Ursprungligt meddelande-----
Fr�n: Luz Berger [mailto:[EMAIL PROTECTED] 
Skickat: den 30 juli 2004 11:21
Till: [EMAIL PROTECTED]
�mne: RE: [WhatsUp Forum] How to know what to look for?


Hi Magnus,

Unfortunately it is not so easy to give an all encompassing view on what has
to be monitored and how to know about it.

First of all, you need to look at your business. Group your infrastructure
along the lines of business. Talk to the line managers on what exactly they
view as absolutely important to have in terms of service. Agree with them on
a SLA ( not easy and certainly not fast to get) or at least derive from
their answers a sort of base line on what part of the infrastructure is
business critical.

Once you have this information, try to map this towards your infrastructure.
This will get you an idea on what you need to look at in terms of systems
(i.e. router, switches, server, etc.). The next step is to define what
parameters to look at for any given device or device group and how to
monitor those. A very important part is to define what sort of reporting is
necessary in order to verify whether a SLA is adhered to or not. Another
important area is the notification (escalation). Again, there is no all size
fits all approach. It heavily depends upon the SLAs and the requirements of
your business. Yes, the size and depth of your IT organisation has an impact
as well.

Only if above has been sufficiently defined and agreed upon can you go to
define the technical aspects of the monitoring and reporting.

Lets take a server as an example. You will need to look at CPU usage, memory
usage, disk subsystem, NICs, etc. How exactly depends upon the role of the
server. Most likely would you like to know about any exception as soon as
possible. Well, this gets you into monitoring events (traps, syslog,
winevent, etc.). One advise, monitor only those items which are necessary in
order to adhere to any SLA. One could monitor everything and then one would
need a lot of manpower and/or systems to make sense out of all those
information.

Now for the bad message. You can not tell in general what sort of mibs and
mib variables to use. It depends upon the make, model and brand of the
device. Everybody does it different. If you use equipment from the big ones
like HP, IBM, Dell, Cisco and the like, things are reasonably easy. All of
them have a lot of info on their web sites and in their documentation. Not
always easy to find, but it is usually there.

A good place to start looking for general SNMP knowledge is
http://www.wtcs.org/snmp4tpc/default.htm.

An excellent set of documentation regarding general system management and
monitoring is here.

Network management
1. Introduction -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap1.pdf
2. Network monitoring -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap2.pdf
3. Network control -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap3.pdf
4. SNMP Network Management Concepts -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap4.pdf
5. ASN.1 notation - only in French
6. SNMP Management Information -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap6.pdf
   SMI: Structure of Management Information 
   MIB: Management Information Base
7. SNMP protocol principles -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap7.pdf
8. RMON basic principles -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap8.pdf
9. RMONv2, SNMPv2, SNMPv3 improvements -
http://www.montefiore.ulg.ac.be/~leduc/cours/ISIR/ISIR-chap9.pdf

This may not be exactly what you have asked for, but this gets you certainly
going into the right direction.

I hope this helps

Luz Berger
Berger Network Consult
http://www.bergerl.com


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Finbom Magnus
Sent: Friday, July 30, 2004 10:05 AM
To: '[EMAIL PROTECTED]'
Subject: [WhatsUp Forum] How to know what to look for?


Hi!

The world of SNMP is quite nice to work with and im learning more every day.

One thing that feels like the heavy part of SNMP is to know what to look
for.

A server has many parts that can break. Both software and hardware. There is
the cpu(and maybe MPU), drives in a raid. Raid-card, several nic's, memory,
power and more...

On the switches there are surely many things that can be monitored as well.

What is the easiest way to find out what things that can be monitored an a
device? The only ways I know this far is to download a MIB, complie and then
browse through it and with help of mibdepot.com find out what every OID is
usefull for.

I dont want to miss anything. Would be boring if I thought of having a good
WUG-config and the a server breaks down becuse I missed that there was that
special OID to monitor..


Best regards
Magnus Finbom
IT-Engineer(Microsoft MCP, MCP+I, MCSE-NT4)
Lansforsakringar Skaraborg
Bank and Insurance
Radhusgatan 8
54129 Skovde
Sweden
phone 0500 77 70 65, gsm 0708 71 70 60, fax 0500 77 70 30
[EMAIL PROTECTED]
http://www.lansforsakringar.se/skaraborg

Please visit http://www.ipswitch.com/support/mailing-lists.html 
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/


Please visit http://www.ipswitch.com/support/mailing-lists.html 
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/

Please visit http://www.ipswitch.com/support/mailing-lists.html 
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/


Please visit http://www.ipswitch.com/support/mailing-lists.html
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/

Reply via email to