Hello. Not to toot my own horn, but take it from someone "who does this sort of thing for a living" /grin/ : To make a long story short... IEEE 802.3 (10Mbps ethernet) distance is not 100ft it is 100m. For buffer, commonly 90m is used. Standard for 802.3z gigabit, as well as 100Mbps ethernet distances are the same for a copper twisted-pair medium. It is not a question of >100m performance will be less; at > 100m it should not be expected to function at all! Collisions have no place (i.e. do not exist) in a full-duplex (switched) environment -- the microsegmentation of a switch alleviates the requirement for the collisioning in CDMA to do rudimentary flow-control. (can you even buy hubs at retail anymore? you'd have to look pretty hard to find one I think -- most Linksys, SMC, DLink, etc SOHO electronics are switching chipsets rather than the simple electrical regeneration of a hub) There are also rules for cascade of hubs and switches (e.g. 3rd cable run must be > 15m to avoid nasty interference). 'nother rule of thumb... if you need to cross AC lines (romex or what have you) do so perpendicularly, and with parallel runs stay at least 6ft away (esp. true when near fluorescent ballasts). The 10ms latency for an ACK as described is not really an element to consider. There is jitter and a 9.6 microsecond interframe gap, perhaps that was what prompted you to recall 10<something> as a risk to traffic. Hopefully this clarifies a few earlier misconceptions. cheers, R. On 16-Nov-05, at 4:18 AM, Garth Phillips wrote: > Everything is always a debate with you isn't it Eric? > > Wireless networks are secure reliable and it is quite > rare to see a greenfield site starting with wired > Ethernet. And the expense difference is not so great, > according to CIO magazine article a few months ago, > about 5% more until you need to make changes to the > infrastructure in which case wireless can be up to 20% > cheaper. > > As for security, both Canada's CSIS and Homeland > Security use wireless as part of their networks. > > On the limit thing, yes there is an ultimate distance > for Ethernet cabling but it is far beyond 100Ft and > VERY unlikely that in a home, even a palatial > 10,000FT-2 one, you'd run up against it. Again the > 100FT run limit is based on a calculation of how much > traffic there is, how many devices and being able to > detect collisions within a specific time limit. this > is why it's called Collision Domain Architecture and > not Limited Distance. While I've long forgotten the > hard formula, I seem to recall that it can not take > more then 10mS for traffic to travel to it's > destination and back to the sending device for > confirmation of a "good send" or collisions are not > seen. The more traffic, the more collisions which slow > down the whole LAN, and the shorter the distances can > be. > > Future proof? While I can't watch television > transmissions over the Internet 'cause the bandwidth > from Rogers Hi-speed isn't enough, I can watch is > seemlessly over a new wireless phone which is far > cheaper than GB ethernet. > > By the way Eric, don't forget that the standard for > both 100MB and GB Ethernet only permit an equally > fixed amount of bandwidth to each device. In 10/100, > the limit is 10MB. I don't what the fixed amount is > for GB. > _________________________________________________ > > For information concerning the MUGLO List just click on > > http://muglo.on.ca/Pages/joinus.html > > Our Archives can be viewed at > > http://www.freelists.org/archives/muglo > > Don't forget to periodically check our web site at: > > http://muglo.on.ca/ _________________________________________________ For information concerning the MUGLO List just click on http://muglo.on.ca/Pages/joinus.html Our Archives can be viewed at http://www.freelists.org/archives/muglo Don't forget to periodically check our web site at: http://muglo.on.ca/