| 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 |
I'm currently working with a customer who is measuring network
utilization using DECmcc. The DECmcc manuals says that the
network utilization is equal to the number of bytes multiplied
by 8 bits divided by the line speed (10Mb/s). However, they
read in an RBMS manual that it is calcualted as the number
of bytes PLUS the frame overhead * 8 /10Mb.
Why the difference? They have compared the MCC numbers with the
data collected by a Data General Network Sniffer. The Sniffer
numbers are higher than the mcc numbers and more in line with
the RBMS numbers. Are the mcc numbers calculated differently?
Is the frame overhead so small that it can be ignored?
Thanks for any comments.
Chris
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 5754.1 | TROOA::BALDOCK | Chris Baldock | Tue Dec 07 1993 08:21 | 20 | |
<<< UPSAR::USER$411:[NOTES$LIBRARY]ETHERNET.NOTE;1 >>>
-< Ethernet Volume 3 >-
================================================================================
Note 393.1 Network Utilization 1 of 1
KONING::KONING "Paul Koning, B-16504" 14 lines 6-DEC-1993 17:14
--------------------------------------------------------------------------------
No, the frame overhead cannot be ignored. Usually there's a substantial
fraction of "small" frames (minimum size, or nearly so).
For 64 byte frames, ignoring the overhead (IPG and preamble) yields a
31% error in the utilization number!
This assumes that the "bytes" used by MCC is the Ethernet frame byte count,
header and trailer included. If they mean the byte counter as reported
in the DNA counters, that counter reports the number of frame INFO bytes,
i.e., 46 for a minimum size packet. If that is used, you get an 82% error.
In short, MCC is wrong.
paul
| |||||