T.R | Title | User | Personal Name | Date | Lines |
---|
945.1 | | NPSS::MDLYONS | Michael D. Lyons DTN 226-6943 | Mon Mar 10 1997 11:03 | 7 |
| ...they are correct - although it has nothing to do with cut
through.
See the keyword head-of-line-blocking for previous discussion on
the topic.
MDL
|
945.2 | technical customers need the GIGAswitch System Overview paper | NPSS::WATERS | I need an egg-laying woolmilkpig. | Thu Mar 13 1997 10:51 | 10 |
| You should supply the GIGAswitch(/FDDI) System Description paper to all
technical customers. There is some discussion of HOLB effect on switch
efficiency, and I thought of the increased efficiency afforded by
hunt groups. (Use a hunt group between the two Gsw/FDDI systems, and
HOLB will be greatly reduced for traffic flowing between switches.)
Since this key paper disappeared from //www.networks.digital.com/ a while
ago, you have to get it directly from the FTP archive of the Digital
Technical Journal at gatekeeper.dec.com/pub/DECinfo/DTJ/ or something
like that. The pointer will of course be in this conference.
|
945.3 | make sure cust isn't distracted by cut-through | NPSS::BENZ | I'm an idiot, and I vote | Thu Mar 13 1997 12:27 | 8 |
| And be sure that the customer isn't focusing on the cut-through
concept, which could lead him to believe that waiting for the token on
a ring could lead to HOLB. That isn't so - the packet goes directly
into the output linecard's buffers. Cut-through is just that it may
start onto the link before the packet finishes getting through the
crossbar.
\chuck
|
945.4 | | AMCFAC::RABAHY | dtn 471-5160, outside 1-810-347-5160 | Fri Mar 14 1997 09:41 | 2 |
| ftp://gatekeeper.dec.com/pub/DEC/DECinfo/DTJ/v6n1/GIGAswitch_System_A_Highperf_0
1jul1994DTJD01SC.txt
|