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

Conference 7.286::fddi

Title:FDDI - The Next Generation
Moderator:NETCAD::STEFANI
Created:Thu Apr 27 1989
Last Modified:Thu Jun 05 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2259
Total number of notes:8590

1680.0. "Defta performance?" by KERNEL::WARDJO () Wed May 10 1995 06:56

    Hello,
    
    Does anyone know if there are any performance problems with the Defta?
    
    A customer has a Defta-aa in a Ultrix 4.4 machine and is seeing a
    considerable decline in performance since he moved over to this. Telnet
    and ftp are much slower (by a factor of 100) than going over the
    ethernet.
    
    There are no errors against the device and the customer has applied a
    patch mentioned in the ULTRIX notes conference, note 4890.
    
    Are there any performance figures for the defta?
    
    Thanks,
    
    Jon
    
T.RTitleUserPersonal
Name
DateLines
1680.1Info from customerKERNEL::WARDJOFri May 12 1995 12:30187
    This is the information the customer has sent me:-
    
       The network map is (its crude....):
    
                         /--------\      193.61.104.x
                         | Sun 10 |------------------|
                         \--------/                  |
                             |                       |
                             |  193.61.107.x         |
                      /----------------\             |
       /--------\    /                  \        /--------\
       | Sun 10 |---|     FDDI Ring     |--------| Router |
       \--------/    \                  /        \--------/
       (mercury)      \----------------/             |
                             |                       |
                             |                       |
                         /--------\      194.66.36.x |
                         |5000/260|------------------|
                         \--------/
                          (rowan)
    
       The FDDI ring has the 5000/260 and 3 Sun 10's on it.
       (All Sun 10's vield similar results)
       I have compiled 'sprayd' from the BSD sources.  All other machines
       come with sprayd installed.
       The arguments to spray are:
    
       -c   - packet count
       -l   - length of packet
       When doing a spray from the Sun 10 to the 5000/260 over the FDDI
       ring we get (typically):
      
    ----------------------------------------------------------------------
       # spray -c 100 rowan-fddi
       sending 100 packets of length 86 to rowan-fddi ...
               74 packets (74.000%) dropped by rowan-fddi
               1 packets/sec, 148 bytes/sec
       # spray -c 100 rowan-fddi
       sending 100 packets of length 86 to rowan-fddi ...
               74 packets (74.000%) dropped by rowan-fddi
               1 packets/sec, 148 bytes/sec
       # spray -c 100 rowan-fddi
       sending 100 packets of length 86 to rowan-fddi ...
               74 packets (74.000%) dropped by rowan-fddi
               1 packets/sec, 148 bytes/sec
      
    ----------------------------------------------------------------------
      (These are three sprays taken simultaneously for comparison)
    
       This happens with packets less than 370 bytes in length:
      
    ----------------------------------------------------------------------
       # spray -c 100 -l 370 rowan-fddi
       sending 100 packets of length 370 to rowan-fddi ...
               72 packets (72.000%) dropped by rowan-fddi
               1 packets/sec, 689 bytes/sec
       # spray -c 100 -l 370 rowan-fddi
       sending 100 packets of length 370 to rowan-fddi ...
               73 packets (73.000%) dropped by rowan-fddi
               1 packets/sec, 664 bytes/sec
       # spray -c 100 -l 370 rowan-fddi
       sending 100 packets of length 370 to rowan-fddi ...
               69 packets (69.000%) dropped by rowan-fddi
               2 packets/sec, 763 bytes/sec
      
    ----------------------------------------------------------------------
       (which seems very odd)
    
       Packets over 371 over the FDDI:
      
    ----------------------------------------------------------------------
       # spray -c 100 -l 371 rowan-fddi
       sending 100 packets of length 374 to rowan-fddi ...
               74 packets (74.000%) dropped by rowan-fddi
               739 packets/sec, 276611 bytes/sec
       # spray -c 100 -l 371 rowan-fddi
       sending 100 packets of length 374 to rowan-fddi ...
               73 packets (73.000%) dropped by rowan-fddi
               768 packets/sec, 287250 bytes/sec
       # spray -c 100 -l 371 rowan-fddi
       sending 100 packets of length 374 to rowan-fddi ...
               75 packets (75.000%) dropped by rowan-fddi
               709 packets/sec, 265489 bytes/sec
      
    ----------------------------------------------------------------------
       Packets of 1K in length, over the FDDI:
      
    ----------------------------------------------------------------------
       # spray -c 100 -l 1024 rowan-fddi
       sending 100 packets of length 1026 to rowan-fddi ...
               64 packets (64.000%) dropped by rowan-fddi
               2 packets/sec, 2456 bytes/sec
       # spray -c 100 -l 1024 rowan-fddi
       sending 100 packets of length 1026 to rowan-fddi ...
               62 packets (62.000%) dropped by rowan-fddi
               2 packets/sec, 2588 bytes/sec
      # spray -c 100 -l 1024 rowan-fddi
       sending 100 packets of length 1026 to rowan-fddi ...
               90 packets (90.000%) dropped by rowan-fddi
               0 packets/sec, 681 bytes/sec
      
    ----------------------------------------------------------------------
       Now if I do a spray from the Sun 10 over the router from the
       104 to the 36 network over the 5000/260's ethernet card we get:
      
    ----------------------------------------------------------------------
       # spray -c 100 194.66.36.2
       sending 100 packets of length 86 to rowan-ether ...
               no packets dropped by rowan-ether
               2560 packets/sec, 220174 bytes/sec
       # spray -c 100 rowan-ether
       sending 100 packets of length 86 to rowan-ether ...
               no packets dropped by rowan-ether
               1828 packets/sec, 157267 bytes/sec
       # spray -c 100 rowan-ether
       sending 100 packets of length 86 to rowan-ether ...
               no packets dropped by rowan-ether
               2133 packets/sec, 183478 bytes/sec
      
    ----------------------------------------------------------------------
       Packets of 1K over the ethernet
      
    ----------------------------------------------------------------------
       # spray -c 100 -l 1024 194.66.36.2
       sending 100 packets of length 1026 to 194.66.36.2 ...
               55 packets (55.000%) dropped by 194.66.36.2
               480 packets/sec, 492511 bytes/sec
       # spray -c 100 -l 1024 194.66.36.2
       sending 100 packets of length 1026 to 194.66.36.2 ...
               61 packets (61.000%) dropped by 194.66.36.2
               369 packets/sec, 379416 bytes/sec
       # spray -c 100 -l 1024 194.66.36.2
       sending 100 packets of length 1026 to 194.66.36.2 ...
               no packets dropped by 194.66.36.2
               1113 packets/sec, 1142055 bytes/sec
      
    ----------------------------------------------------------------------
    
    
       If we do a spray from one Sun 10 to another over the fddi we get:
      
    ----------------------------------------------------------------------
       # spray -c 100  mercury
       sending 100 packets of length 86 to mercury ...
               no packets dropped by mercury
               3025 packets/sec, 260188 bytes/sec
       # spray -c 100  mercury
       sending 100 packets of length 86 to mercury ...
               no packets dropped by mercury
               3177 packets/sec, 273302 bytes/sec
       # spray -c 100  mercury
       sending 100 packets of length 86 to mercury ...
               no packets dropped by mercury
               3248 packets/sec, 279329 bytes/sec
      
    ----------------------------------------------------------------------
       And 1K packets:
      
    ----------------------------------------------------------------------
       # spray -c 100  -l 1024 mercury
       sending 100 packets of length 1026 to mercury ...
               37 packets (37.000%) dropped by mercury
               1607 packets/sec, 1649686 bytes/sec
       # spray -c 100  -l 1024 mercury
       sending 100 packets of length 1026 to mercury ...
               37 packets (37.000%) dropped by mercury
               1607 packets/sec, 1649686 bytes/sec
       # spray -c 100  -l 1024 mercury
       sending 100 packets of length 1026 to mercury ...
               44 packets (44.000%) dropped by mercury
               1573 packets/sec, 1614522 bytes/sec
       # spray -c 100  -l 1024 mercury
       sending 100 packets of length 1026 to mercury ...
               22 packets (22.000%) dropped by mercury
               1747 packets/sec, 1792661 bytes/sec
      
    ----------------------------------------------------------------------
    
       I know 'spray' really isnt a precise tool, but we are seeing
    problems using telnet and ftp .....
--------------------------------------------------------------------------------
    
    Can anyone help with this.
    
    Regards,
    
    Jon