Curiosity killed the cat
		Date: 09/03/07 
		(IT Professionals)    Keywords: no keywords
Perhaps one of you can shed some light on this for me.  It's not exactly a problem, but just a mystery that doesn't make sense to me.
Say you have a dedicated box being used as a Ghostcast server. Said box has a gigabit NIC card in it.  If I plug one end of a 3ft Cat6 patch cable into this box, and the other into a similar computer with an identical NIC that I want to deploy an image to, I get a Ghostcasting speed of approximately 500MBytes/min.
We have a Cisco gigabit switch available to use (a Catalyst 5000G, specifically).  If I plug 6 computers into the switch (including the server) and assign the same IP/Subnet range to all of them, I get a Ghostcasting speed of nearly 2000MBytes/min.  All cables are Cat6 and approximately the same length, and I'm spitting out exactly the same image.
I suspect that the speed difference is due to my using a patch cable and not a crossover; I'm aware that many cards will automatically correct this through hardware - would that slow it down?  What does the switch do differently that speeds it up so dramatically?  Why is that that plugging a short Cat6 cable directly between server and client is four times slower than 6 computers running through a gig switch? Thank you for any insight you provide.
~Elliot
Source: http://community.livejournal.com/itprofessionals/64207.html