DComTalk.com

I've got a network question for you all, hope you can help. I'm
designing a network, and I have four buildings that I need to get
connectivity to. The buildings are laid out in a sort of upside-down U
shape, with the closest building to our main switches about 600ft away.
We've been able to get connectivity to this building by running 100m of
cat5, putting a switch down in the middle, then continuing on.

The other buildings are presenting somewhat of a challenge, however.
Ideally, I'd like to plunk down a central switch in a fifth building
that's centrally placed relative to all of these, but wouldn't you know
- that building may not be there much longer. So, the only feasible way
I can see to get connectivity to my buildings is to daisy-chain them
all the way around - run wire to a building, put a switch in it for the
local users, and run another wire off that switch to the next building.
The distances between the buildings are long, but they're well under
the max recommended Cat5 lengths.

So, my questions are: Is this workable, and will the guys at the very
end of the daisy-chain (which will be at the end of about 1400' of
cable passing through 4 intermediary switches) have any connectivity to
speak of?

Thanks!

One word: fiber! Otherwise any nearby lightning strikes will
induce ground loops and fry equipment. You will be even more
vulnerable with a daisy-chain setup. Any link can fail.

For little more than the price of more fibre you can pull cable in a
daisychain duct system but use just one switch in a central location
to make a star LAN configuration which is always preferable.

If that isn't clear, here's an attempt at a diagram. The pipe from the
switch to bldg1 has 4 (really 8) fibers, the next pipe hop has three,
and so on.


S---------- bldg1
W------------------------bldg2
I-----------------------------------bldg3
T------------------------------------------------bldg4
C
H

This way you wan't need bicycles and walki talkis to troubleshoot your
network and the users in building 4 will get the same reliability and
performance as the users next to the switch.


--
a d y k e s @ p a n i x . c o m

Don't blame me. I voted for Gore.

I've got a network question for you all, hope you can help. I'm
designing a network, and I have four buildings that I need to get
connectivity to. The buildings are laid out in a sort of upside-down U
shape, with the closest building to our main switches about 600ft away.

One word: fiber! Otherwise any nearby lightning strikes will
induce ground loops and fry equipment. You will be even more
vulnerable with a daisy-chain setup. Any link can fail.

-- Robert

So, my questions are: Is this workable, and will the guys at the very
end of the daisy-chain (which will be at the end of about 1400' of
cable passing through 4 intermediary switches) have any connectivity to
speak of?

Other than a bit of latency, there shouldn't be much difference. However,
why not just run fibre? It can support far greater bandwidth over greater
distances. This way, you could run cables directly from the various
locations to a central point.

Robert Redelmeier wrote:

One word: fiber! Otherwise any nearby lightning strikes will
induce ground loops and fry equipment. You will be even more
vulnerable with a daisy-chain setup. Any link can fail.

Lightning doesn't induce ground loops. Ground loops are created when cables
are grounded at more than one point. UTP cables aren't grounded anywhere.

Lightning doesn't induce ground loops. Ground loops are
created when cables are grounded at more than one point.
UTP cables aren't grounded anywhere.

Sorry for my terminology which is a little short and imprecise.
The main mechanism for lightening damage (nearby, not direct
strikes) indeed involves ground loops imposed by the strike:

Lightening hits at AC voltages of several million volts. It does
not dissipate immediately into the ground but imposes a potential
gradient of several thousand volts per meter. Two different
buildings groundstakes go to very different potentials and
harmlessly lift all the chassis on premises to these potentials.
However, the UTP transceiver transformers only offer a combined
thousand volts or so of isolation. They breakdown and a ground
loop flows.

As a numerical example, image a lightning strike 350m away from
one building (groundstake) and 400m away from another. At this
disance and soil conditions (uniform, ha!) assume the ground
gradient is only 1,000 V/m. But those two buildings go to 50
kV ground differential during the strike. The UTP will fry
transformers and electronics.


-- Robert

I've never attempted it but I think there will be significant latency. Definitely go with fiber. I wouldn't even risk it.