h a l f b a k e r yTip your server.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Token Ring networks work real well and were fairly stable. They didn't catch on for several reasons. What a token network does is send a little token around and when a computer receives the token it is able to send or receive information across the network. One problem with this was the size of the
package that was able to be sent through the network. Although it was small, it was stable and didn't have collisions and need to be resent, as in an ethernet network. Another reason, and from what I hear was a large reason, that this type of network make it was that IBM had the rights to it. Since IBM had it, they wanted money for it, so to use it you had to pay IBM for the tok adapters, for the wires attached to the adapters, for the drivers, etc etc etc.
Ethernet networks aren't as stable as token. In this environment all the computers on the network talk at the same time. If the data got across good, if not, oh well try again. If the data from one computer ran into the data of another computer's, there would be a collision and the data wouldn’t reach the destination. So collision detection was devised and put into place. What this does is make is so that if a collision was detected a signal would be sent back to the sending computers where a formula would be figured out and the data would be sent back across the network only delayed a few nanoseconds (the number of nanoseconds devised by the formula).
This idea is about merging these two types of networks. Have a LAN that has several tokens going around. The tokens are able to jump each other so a backup of information would be avoided. Also, as soon as a token attached itself to a computer the whole queue would be sent and not just a package or window of information. If two or more computers get a token at the same time and try to send their information, a collision detection device would send out a signal to the tokenized machines and a formula would be worked out and the data sent back across the network delayed by a few nanoseconds devised by the formula. This type of LAN would be stable from the token side and faster because of the ethernet side.
[link]
|
| |
No, more like "Monopoly" meets "Risk" ..... "Destroy all Blue Armies"....... |
|
| |
barnzenen: I think you've been overtaken by technology. Collisions only occur on the same segment of an ethernet network. Use a switch and every port is its own segment. No more collisions. |
|
| |
One big problem with a token-ring network is that either the computers have to be physically wired in a "ring" configuration, or the computers have to collectively know the identity of everybody on the network. Ethernet imposes no such restrictions: computers may be added to the network without any other computer needing to have any knowledge about them. |
|
| |
One of the primary advantages of the token network is that when a large number of nodes want to send a large amount of data, the even distribution of availabiliity windows means that each node gets an even chance. If one node has a constant stream of data, they can collide and consequently jam the network that you are describing. Another problem would emerge when you wanted to send data onto a larger area network. You would need to double address the packets to be passed through a selective router and on to the larger token network. |
|
| |
It sounds to me what is being proposed serves the same purpose as a regular Ethernet network, but with more complexity. There are still multiple computers trying to talk on the network at the same time, and there still are collisions. Yet only a limited number of computers can comunicate at one time over the entire network So I fail to see the advantage.
It sounds to me what is being proposed serves the same purpose as a regular Ethernet network, but with more complexity. There are still multiple computers trying to talk on the network at the same time, and there still are collisions. Yet only a limited number of computers can communicate at one time over the entire network So I fail to see the advantage.
On another note, contrary to what was said an earlier post. Token-Ring networks only need to be logically (virtually) wired in a "ring" configuration (my understanding is that Token-Ring hubs manage this internally so no special configuration is needed on the networked PCs). Physically they can be wired in a "star" (several PCs connected trough a hub) configuration |
|
| |
I don't think you could play Doom on a token LAN. That's why it failed. :-) |
|
| |
You want to take the caps off how much you can send before you release the token, as I understand it, but the problem is then Computer A wants to send a 10 gig file to Computer B. Computer C wants to send a 56 byte ACK to Computer D. C has to wait for A to send the entire file before it can send its tiny, little thing. That's why there are caps on that. You want to fix it by adding more tokens, but the problem is still there. It's less likely, but now A wants to send 10 gigs to B, D wants to send 5 gigs to B, C wants to ACK something, and both tokens are busy elsewhere. The way to solve that is to make as many tokens as there are computers in the network and add collision detection, and you have ethernet. |
|
| |
I don't see the benefit. Still could do it if you felt like it, though. |
|
| |