[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

210.0. "Max stations, RFC 791, and ANSI SMT spec's" by RIPPLE::KOPEC_ST (Seattle DNT/Sales - DTN 545-4207) Sun Feb 24 1991 21:13

1)  When we state that we will support up to 500 stations on 1 FDDI LAN, I 
    assume we are talking about end nodes and bridges only, *not* concentrators.
    In other words, from a present day (CY91) DEC h/w point of view... we are 
    talking only about DECstation 5000s, VAXes with XMI buses, and the 
    DECbridge 500.
	a)  Is that correct?
	b)  What is the technical reason(s) for the limit?


2)  I have read in a number of places that DECbridge 500 "supports the IP 
    packet fragmentation per RFC 791 to handle differences in maximum packet 
    sizes between FDDI and Ethernet LANs without performance degradation."

    The implication is that the protocol being passed from FDDI<->Ethernet
    *must* be IP in order for this feature to kick in.  

    a) Is this true?
    b) What about DECnet Phase IV today, SCA today, or DECnet Phase V tomorrow?


3)  In the new FDDI slide set on OPAL, one of the slides states that the yet 
    to be product announced XMI controller will support "ANSI FDDI SMT 
    capabilities for standards-compliant station management".  

    a)  Could someone list these ANSI approved SMT capabilities?
    b)  Is it correct to assume that the FDDIcontroller 700 has these same 
	capabilities?
T.RTitleUserPersonal
Name
DateLines
210.1For SCA, fragmentation at hostHYEND::KPARRISDECclustersMon Feb 25 1991 08:2312
I'll tackle a small piece:

2)   b) What about DECnet Phase IV today, SCA today, or DECnet Phase V tomorrow?

For SCA, fragmentation is handled at the host level.  The driver detects
whether there is an Ethernet segment anywhere along the path from another node
by examining the priority field in the frame control byte.  The default value
is 4, and our 10/100 bridges set it to zero (the IEEE is reportedly expected to
include this in some specification so this behavior will become standard). 

If there are no 10/100 bridges in the path, full-size FDDI packets can be used,
otherwise only Ethernet-size packets are sent. 
210.2Some more answersJUMP4::JOYGet a life!Mon Feb 25 1991 08:3217
    Stan,
       1) Concentrators ARE included in the 500 station maximum per ring.
    What you're really counting is the number of MACs in the ring. 500 is
    the theortical limit. 200-250 is the practical limit due to the latency
    incurred in the ring each time the token/packets pass thru a station.
    
    2) There are several TCP protocols that the bridge will fragment today,
    IP being the most common (EGP and UDP are two others). THe
    fragmentation is done based on RFC 791.
    
    For DECnet, see the ENUF::PHASEV notesfile for how OSI will handle
    fragmentation (also mentioned in an earlier reply in this file
    somewhere.
    
    
    Debbie
    
210.3slight clarificationTOOK::ROSENBAUMRich Rosenbaum, TaN/OSF, 226-5922Mon Feb 25 1991 12:1010
    re: -1
    2) There are several TCP protocols that the bridge will fragment today,
    IP being the most common (EGP and UDP are two others). THe
    fragmentation is done based on RFC 791.
    
      EGP and UDP (and TCP, for that matter) are layered on IP, so this
      is all really just one kind of fragmentation (RFC 791 is the IP RFC).
    
    Rich