Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU |
I tend to believe that most units now being manufactured use chipsets using the "N-way" autosensing features that allow for maximim throughput (including full duplex operation) as determined by the slowest junction (and, of course, any extraneous line noise). As long as each connection has that feature enabled I expect that they will work together to provide the maximum throughput for the 2 devices connected at any one point in time (older 10baseT NICs may not). So check the documentation for each of your NICs and any new hub or switch you choose to purchase. I know that the LinkSys 100TX boards are both inexpensive, some offer features such as Wake on LAN and are Linux compatible (at least with the later SuSE and kernels supporting Tulip chips). A switch obviously has benefits if you have a lot of devices sharing (unconflicted) at the same time, and the small price differential argues that a 100TX switch is now a better investment. I am curious if you are running a mixed population of NICs (10 & 100 Mbs) whether that conflict would impede the speed and efficiency of a 100TX based server connection (I had once and only once read this was a possibility for a switch but no one else has ever confirmed the possibility). That might be the difference between choosing a switch or a hub. You may also find some advantage segregating your 10 base systems from 100TX capable units by keeping your slower units on the 10BaseT hub and using a 100TX switch to link the faster units while leaving a single channel to the slower hub and its nodes. I have a SoHo branded 10 base hub that has given me problems (particularly but not exclusively) using my Macs on it but SMC and Asante have some good deals on both hubs and switches (and work with my various Macs and Linux PCs). They along with most brands should work just fine for Linux. One thing that might offer efficiency advantages for mixed speed networks is a larger memory buffer. More expensive switches offer larger buffers and so should have better throughput on mixed networks. I believe that is the big difference between the Asante 4000 "P" and "E" series of switches and is probably true of most brands. Jerry Feldman wrote: > Just a question. I plan to upgrade my home network to 100 baseT which will > require a new hub, and at least 1 new NIC. The question is can a 100Mbit > hub support 100Mbits between 2 100Mbit NICS when one or both of the other > hub ports are connected to a 10Mb NIC. Or will just the connection to a > 10Mbit NIC cause the entire network to run at 10. The issue is that the > Alpha has an on-board 10Mbit NIC and another computer is idle most of the > time. I know I can do this with a switch, but that is an overkill. > -- > Jerry Feldman <gaf at blu.org> > Boston Linux and Unix user group > http://www.blu.org > > - > Subcription/unsubscription/info requests: send e-mail with > "subscribe", "unsubscribe", or "info" on the first line of the > message body to discuss-request at blu.org (Subject line is ignored). - Subcription/unsubscription/info requests: send e-mail with "subscribe", "unsubscribe", or "info" on the first line of the message body to discuss-request at blu.org (Subject line is ignored).
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |