[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference noted::netview-nt

Title:Notes Conference for TME 10 NetView for NT NT
Moderator:TUXEDO::BAKER
Created:Mon Apr 03 1995
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:540
Total number of notes:1903

527.0. "Data collection question on octets" by NETRIX::"penno@mail.dec.com" (Murray Penno) Mon Mar 10 1997 19:28

Hi,

I have a customer who is using PNV to do some data collection from
Bay routers. He is collecting from the mib variable ifinoctects and
ifoutoctets so that he can work out utilisation.

The description for the mib variable is total number of octets
received on the interface.  The customer seems to think that in fact
it is octets per second.  Can anyone explain this?

Also with compression on the routers is the number of octets before
or after compression. (ie the serial side or the ethernet side)

Any help appreciated

Thanks

Murray Penno
[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name		DateLines
527.1SMURF::DANIELETue Mar 11 1997 13:0918
>The description for the mib variable is total number of octets
>received on the interface.  The customer seems to think that in fact
>it is octets per second.  Can anyone explain this?

Yes.  The customer is wrong :-)

It is a counter, not a rate.  Applications typically sample this counter
and sysUpTime, then calculate utilization.  (NetView probably can do this
for you, I'm not a NetView person.)

>Also with compression on the routers is the number of octets before
>or after compression. (ie the serial side or the ethernet side)

I don't really understand this question, but would think ifInOctets counts 
octets received at the interface, regardless of whether or not a higher layer
will compress or expand them.

Mike
527.2Data already normalised?WOTVAX::dhcp20.olo.dec.com::hunt_jsTue Mar 11 1997 15:2124
In my experience...

I agree the ifinoctets and ifoutoctets, as presented by NETVIEW, is on a per 
second basis.

Thus you do not need to normalise data across any time period as NETVIEW 
appears to be doing it for you.

Thus to get line utilisation just: ifinoctets * 8 / linespeed * 100

Its easy to prove this. The data would be an increasing number if it were'nt 
already normalised for you. Also if you collect "bits" instead of "Octets" and 
divide by the line speed you get the same answer as octets*8/linespeed.

Something also to be aware of...

If you use the standard % UTILISATION data gathering mechanism 
(bandwidthutilhd) then beware as this is taking both "in" data AND "out" data 
but only dividing by the line speed. Thus the resulting data is out of 200% 
not 100%!

Hope this helps.

Cheers, Jim
527.3SMURF::DANIELEWed Mar 12 1997 13:0014
>In my experience...

>I agree the ifinoctets and ifoutoctets, as presented by NETVIEW, is on a per 
>second basis.

I hope I haven't confused the issue.  I was referring to the values returned
by a conformant SNMP agent.  These values (as displayed by a MIB browser)
are counters, not rates.  They always increase until they wrap to 0, or 
some discontinuity occurs (interface/agent go down).

If NetView has some other display/function that automatically does sampling
and rate calculation, that's beyond my knowledge.

Mike