| Title: | DECmcc user notes file. Does not replace IPMT. |
| Notice: | Use IPMT for problems. Newsletter location in note 6187 |
| Moderator: | TAEC::BEROUD |
| Created: | Mon Aug 21 1989 |
| Last Modified: | Wed Jun 04 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6497 |
| Total number of notes: | 27359 |
Cross posted in the vitalink notesfile...
<<< IAMOK::APL01:[NOTES$LIBRARY]VITALINK_BRIDGE.NOTE;1 >>>
-< Vitalink Bridges >-
================================================================================
Note 839.0 utilization via snmp on 350 t1 No replies
HEN::SCHNEIDER "Rick Schneider DTN 256-5814 SYO" 28 lines 2-MAY-1994 19:08
--------------------------------------------------------------------------------
I'm attempting to look at line utilizations on a Vitalink Network.
I've setup DECmcc to poll several of the bridges to collect statistics
on the lines of interest.
The BIT Rate for each line reflects the actual modem speed per line
(ie. 56K line = bit rate of 56000, t1 lines = bit rate of 1344000)
The export polling rate is every 5 minutes (bridges were rebooted
april 11, 1994). Sample period started april 23 through the 29th...
Using MCC reports to generate the hourly utilization rate reveals
frequent utilizations greater than 500%(inbound and outbound).
Is there something I missed, or is it a matter of garbage into mcc
gets garbage out of mcc?
I dumped the sql database and worked through the calculations provided
in the performance manual appendix for MCC. The number coorespond,
although I get different numbers for IFINOCTETS and RCVD_OCTETS (
similar differences with the xmitd_octets and ifoutoctets).
The TRANSLAN of interest is a 350 running 10.5.1 software. Would newer
software help?
thanks!
rick
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 5973.1 | a starting point... | CSOADM::ROTH | What, me worry? | Wed May 04 1994 11:06 | 8 |
Do the following: $ man/ent show snmp <vitalink name here> interface * ifSpeed Do all of the speeds look correct? Lee | |||||
| 5973.2 | yep they are correct... | HEN::SCHNEIDER | Rick Schneider DTN 256-5814 SYO | Fri May 06 1994 15:00 | 4 |
Yes, they all are correct 56000 for 56k and 1344000 for the T1 lines
rick
| |||||
| 5973.3 | CSOADM::ROTH | What, me worry? | Mon May 09 1994 15:10 | 3 | |
I give up... I dunno. Lee | |||||
| 5973.4 | after further review... | HEN::SCHNEIDER | Rick Schneider DTN 256-5814 SYO | Mon May 09 1994 19:57 | 24 |
After looking at the data more thoroughly, I do see speed values that
are NULL, with valid counters for IFINOCTETS, IFOUTOCTETS, etc...
The bogus utilization occurs more often than NULL speed values in the
database. I could see how a Null value might affect the calculation for
the interval which contains the sample with a NULL speed, but not
widespread basis i'm getting... only a few seemingly correct values for
a 24 hour sample.
One other issue that might cause bogus utilization, is that the
customer NEVER zeros out the counters, so they tend to wrap every
couple days on the T1 circuit. I doubt MCC can handle it since it can't
know the counter size of every vendors equipment... Vitalink TransLAN
350s wraps at 2^32...
How does MCC calculate the utilization? I'm sampling every 5 minutes
and reporting via the hourly graph from MCC... Will MCC take the first
and last sample of the time period and subtract the counters to
determine the octets sent/received? Or does it use some other method?
thanks,
rick
| |||||