<Reginald, it sounds like you are challenging me. Why the fierce attitude? If you don't understand networking, how do you hope to argue it with me?>
Glad to see we are getting a bit more polite in our responses. And to clarify, there is a distinct difference between asking one to support there assertions and "challenging" someone. I have no "fierce attitude". I openly admitted that I would not challenge you on networking, but would challenge you on logic if I felt it necessary.
Now back to your enlightening post. We (at least I) are now getting a clearer picture. Lets try and examine this situation in the context of a real world appliation of the NC.
<It all depends on topology, Reginald. 50 Users on ethernet? The topology would probably be multiple hubs of 8-10 users each connected to a fast ethernet backbone.>
My example stated that there were 150 desktops on an ethernet, in which there were 50 users on it simultaneuosly at any given moment. So wouldn't the load scenario be a bit heavier than you described, especailly if it scales exponentially?
< If 10 users were using the file server at their 10Mbps limits, they would saturate that file server (network efficiency would begin degrading exponentially, not linearly). However, in typical scenerios, out of 50 users you would probably never see 10 users hit a file server at 10Mpbs each. You would probably never even see more than 1 or 2 at a time do this.>
Taking into cosideration that usage in the proposed scenario is three times greater than you described, coupled with the fact that the traffic on this ethernet would be SIGNIFICANTLY increased by the use of the NC thin clients (which may not be able to cache but so much locally and do not store any data or binaries locally - therefore must move all of this back and forth along the ethernet pipes creating SIGNIFICANTLY more traffic than would normally be expected from traditional "fat" PCs), isn't it feasible, if not likely, that 10 users out of 150 would hit a file server at 10 Mbps each through the course of a business week (remember the network will be moving databases, spreadsheets, applications and the like to and from each and every user). And remeber that this is a relatively difficult question to answer correctly due to the fact that full scale deploymetn of NCs have yet to be tested in a real working environment, or am I mistaken? Most answers are conjecture or educated assumptions at best, which leads to my assertions that the claims for NCs are currently bogus when costs are compared to the PC.
<A lone T1? 1.54Mbps? It is no comparison whatsoever to the topology I just covered. In the above, every one of those 50 users has an effective bandwidth of 10Mbps. The math is more complicated than you think - go take a networking course if you want to understand this stuff a little better.>
Each and every one of those 50 users may have a theoretical bandwith of 10Mbps, but what about the connection and possible saturation at the server. My example stated that a lone user on the T1 would also have solo access to the server. It is the server that is most likely to be saturated isn't it?
And finally does the average corporation have a 100Mbps ethernet?
Remember, have manners when you reply or you will make me mad at you. |
