[rampel]

I have a cable modem connection which is 1.5Mbps (can go upto 3 Mbps if I pay more), connected to my 9600 Mac's inbuilt Ethernet port.

Question: will I benefit (see any speed increase) if I install a PCI Ethernet card capable of 1 Gig ( 10/100/1000 Mbps)?

Which one in the chain is the slowest and determines the transfer speed?

[DeathStrike]

there would be no speed increase by doing that

a cable modem is a 10megabit device. Me having my cable modem hooked to a 100megabit port doesent make a different either.

[rampel]

Originally posted by DeathStrike:
there would be no speed increase by doing that

a cable modem is a 10megabit device. Me having my cable modem hooked to a 100megabit port doesent make a different either.

Thanks, useful to know

[BkueKanoodle]

The bottle neck would be the cable connection, which tops out at 1.5 mb so you'd be wasting 998.5 mb of extra capacity with a gigabit card.

[ghporter]

Yep, your cable connection (not just the modem) is the slowest link in the chain.

You could see a benefit within a LAN if all of the computers had faster NICs, but that would only affect file transfers between computers. Right now just about every broadband connection runs at or below the 10Mbps theoretical rate of 10BaseT ethernet, so small office/home networks really don't benefit from gigabit ethernet-except for file transfers within the LAN.

[rampel]

Originally posted by GHPorter:
Yep, your cable connection (not just the modem) is the slowest link in the chain.

You could see a benefit within a LAN if all of the computers had faster NICs, but that would only affect file transfers between computers. Right now just about every broadband connection runs at or below the 10Mbps theoretical rate of 10BaseT ethernet, so small office/home networks really don't benefit from gigabit ethernet-except for file transfers within the LAN.

Thanks, I get the full picture now

[ghporter]

Originally posted by rampel:
Thanks, I get the full picture now

I hope my reply didn't sound like beating you over the head with facts and figures. It's just that so often we answer questions with the least number of words necessary to answer the surface question, ignoring the fundamental question (usually "why?") that prompted the question that was asked. I like to answer the whole question, so others browsing the board can pick up something more than a yes or no answer.

I sometimes have trouble at parties because of this habit...

[rampel]

Originally posted by GHPorter:
I hope my reply didn't sound like beating you over the head with facts and figures. It's just that so often we answer questions with the least number of words necessary to answer the surface question, ignoring the fundamental question (usually "why?") that prompted the question that was asked. I like to answer the whole question, so others browsing the board can pick up something more than a yes or no answer.

I sometimes have trouble at parties because of this habit...

Not at all, not to warry. On the contrary, you really made the picture complete for me. Thanks

[selowitch]

Great discussion, very interesting and topical!

I wonder if sometime in the near future we might see broadband connections that would be of sufficient speed to merit greater investment in gigabit Ethernet.

Question: When we say 10BT, 100BT, how does that translate/convert to megabits per second (Mbps)?

[ghporter]

The "10" or "100" is the theoretical maximum data transfer rate for these connections in megabits per second. This includes the overhead management data required to operate and monitor the link.

[selowitch]

Originally posted by GHPorter:
The "10" or "100" is the theoretical maximum data transfer rate for these connections in mega<u>bits</u> per second. This includes the overhead management data required to operate and monitor the link.

I guess that means that the only time we'd ever want to consider upgrading to 1000BT for the sake of an Internet connection is if we were using a T-1 or some variant thereof?

[ghporter]

Actually, a T1 line is almost identical in speed to a standard 1.5Mbps DSL connection. This is much too slow to warrant anything faster than 10BT.

[tooki]

Originally posted by selowitch:
I guess that means that the only time we'd ever want to consider upgrading to 1000BT for the sake of an Internet connection is if we were using a T-1 or some variant thereof?

There is no internet connection available to end-users that comes close.

A T-1 is 1.544Mbps, a T-3 is 45Mbps -- but a T-1 costs between $500-1000 a month. A T-3 is much, much, much more expensive still. The really fast lines -- OC-3s for example (3Gbps) and faster cost tens or hundreds of thousands of dollars a month.

Even so, you wouldn't benefit, because the Internet's structure limits effective speed of a given connection. Faster lines don't end up speeding up a single download, but rather allow many more downloads to occur simultaneously. In any case, it's unlikely any single computer could benefit from having a faster ethernet card.

tooki

[kampl]

Originally posted by tooki:
There is no internet connection available to end-users that comes close.

A T-1 is 1.544Mbps, a T-3 is 45Mbps -- but a T-1 costs between $500-1000 a month. A T-3 is much, much, much more expensive still. The really fast lines -- OC-3s for example (3Gbps) and faster cost tens or hundreds of thousands of dollars a month.

Even so, you wouldn't benefit, because the Internet's structure limits effective speed of a given connection. Faster lines don't end up speeding up a single download, but rather allow many more downloads to occur simultaneously. In any case, it's unlikely any single computer could benefit from having a faster ethernet card.

tooki

OC-3 is much slower than that, roughly 155Mbps. My ATM core runs at these speeds on the MAN. A pair of machines running 100Mb full duplex can saturate a link of this type (seen it happen) depending on what is being done. For a brief run down of optical carrier speeds see this link .