[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: My network speed is only 10MB



On 2/4/2012 10:03 AM, Camaleón wrote:
> On Sat, 04 Feb 2012 08:47:16 -0600, Stan Hoeppner wrote:
> 
>> On 2/4/2012 6:53 AM, Camaleón wrote:
>>
>>> No, I can't see why is not that popular within the US, there are many
>>> advantadges for having shielded cables because external interferences -
>>> that are not always under your control- still apply (e.g., wireless
>>> connections, proximity to high power lines or electrical equipments...)
>>> whose effects can be properly minimized with cabling shielding.
>>
>> These same effects are eliminated, in the US, by installation standards.
>>  These are are always under the control of the installer/contractor.
>> There is no guess work involved and everything is predictable and "under
>> your control".
> 
> You never know what kind of company is going to be installed next to your 
> garden, right? So one day you open the door and find a power plant is 
> your brand-new neighbor. At the time you (or a professional contractor) 
> installed the network facilities neither of you could guess the new 
> situation and the installation was not prepared to deal with that. 
> 
> You can control what you know and prepare for unforeseen upcoming 
> circumstances is always a good strategy.

You have a fundamental misunderstanding induced EMI/RFI.  The source of
the interference must be relatively close, physically, to the cable, in
order for the cable to pick up sufficient noise to interfere with
signals.  A power plant, or even a Tesla coil, in the building next door
will have zero effect on your cables' signal integrity.

>> Wireless signals do not affect UTP ethernet cabling.  Proximity to
>> internal AC cabling, induction sources such as power distribution
>> closets, AC motors, florescent lights, etc, is avoided during building
>> construction or retrofit, because the cable plant is included in the
>> architectural design, just like water pipes, sewer pipes, electrical
>> conduit, etc.
> 
> Not at all. This is something you can't decide. Only when you start 
> having problems you do figure something has changed and start looking 
> outside your own premises and search for the root of the problem. 

I've done structured cabling, professionally.  I assure you it can be
done the way I describe above, as that is the way it's done in the
States every single day.  Standards (or lack of following them) must be
different in your locale.

>> In the US, in the case of environments such as manufacturing floors etc
>> with horrific EMI levels, fiber is used instead of UTP CAT5/6.  With EFI
>> levels that high, even STP won't save you.
> 
> Fiber is (still) not an option for the LAN. 

You facilities must be relatively small there in Spain.  In the US
closet stacks are often more than 100 meters from central switching so
fiber is the only option for connecting the closets, regardless of
whether copper or fiber is run from the closets to the cubicles/offices.
 This is the case in almost all multiple floor office buildings where
the datacenter is most often on the ground floor or in the basement.

> Here in Spain, the ISPs have just started to install FTTH for residential 
> and business users just a year or so ago, but it's not widely implemented 
> and it's only for Internet access. Cooper is the king here for the LAN 
> environment.

I would guess you simply have a shortage of trained fiber installers in
Spain.  Every major city in the US has multiple electrical contractors
with trained and certified IBEW and on-union fiber installers on staff.

>> In summary, when installed correctly, UTP ethernet cable is superior to
>> STP, due to the lower cost of the cable, connectors, and patch panels,
>> and labor.
> 
> I disagree. Most of the UTP cables are "user-made", badly assembled, poor 
> quality and installed completely unstested while shielded cables came 
> certified from the manufacturer.

Again, you obviously have a lack of qualified cable installers there,
lack of testing equipment, or poor contracts.  Installation contracts
here dictate that each cable run is tested at the patch panel on both
ends and a report generated for each run showing it meets specification.
 One such common tool in wide use in the US:

http://www.flukenetworks.com/datacom-cabling/copper-testing/dtx-cableanalyzer-series

Tools costing less and many times more are used here.  This is but one
example.

>> Something worth mentioning is that over the past 10-12 years, a high
>> percentage of new large office buildings and high rise apartments
>> constructed in the US have had 100% fiber cable plants, no copper
>> whatsoever, even including fiber into the cubicles.  The only copper UTP
>> in such facilities being patch cables from servers to core switches and
>> between switch stacks.  Note that datacenter copper is not part of the
>> cable plant.  "Cable plant" is the slang term for "structural cable
>> installation", plant used as a verb, analogous to planting tomatoes in
>> one's garden.
> 
> Fiber is another different thing. We do also have it installed since the 
> last summer (4 FTTH lines, a 16-fibers cable) but working with the fiber 
> can be only done by certified installers and the required tools are very 
> expensive, not every company can afford that.

Again, nearly every structured cable contractor here in the states does
both copper and fiber.  Fiber is mundane here.  You refer to fiber as
something used only to connect providers to a customer premise,
something special, ethereal.  Here fiber is used just as often, if not
more often, inside buildings as it is between sites.  This should not be
surprising considering the US leads the world in optical fiber consumption.

>> Cost is one of the drivers.  Today a 1000ft spool of 62.5/125 multimode
>> fiber is equivalent or cheaper than CAT6a UTP.  The installation labor
>> is about the same as UTP.  We'll continue to see more fiber to the
>> desktop as the cost of copper continues to increase.  The switch cost in
>> an all fiber plant is higher per port due to the multimode transceivers,
>> but not prohibitively so when purchased in volume.
> 
> The price of the fiber devices... Routers, NIC adapters, switches, etc... 
> they're still very costly (almost prohibitive here, but Spain is 15 years 
> ahead USA in these regards) and unless you have an infrastructure capable 
> of making use of such speeds you're wasting your money.

You mean "behind" the US, I think.  The first all fiber cable plant I
was personally aware of was the new MasterCard facility in St. Louis.
That facility was built 12 years ago, with fiber to the desktop at
100FDX data rate.  With a facility of that size and the number of ports
required, the additional cost of fiber cable and transceiver hardware
was pretty miniscule compared to the labor cost of the cable plant.

Going with fiber insured future proofing.  62.5um FDDI grade cable can
support 10GbE over 300m using 10GBASE-LX4.  So that fiber installed 12
years ago can support 10GbE to the desktop today, if required/desired.
It's kinda tough to achieve 10GbE over copper at 300m, let alone 100m.
They've already converted to GbE at the desktop.  And the big win, from
day one, is the installers didn't have to be concerned about EMI/RFI
during the design and installation, yielding more flexibility in cable
routing, which did save some labor and materials cost to boot.

Thus I disagree with your "wasting money" comment.  At the time the
above cable plant was performed fiber had no speed advantage over CAT5.
 It did have other advantages, as I mentioned, and still does today.  At
least here in the US.  Labor must be really cheap in Spain if the cost
of fiber cable and transceiver hardware is the deciding cost factor.
Given your assertion that installers there are not qualified and do
shoddy cable plant work, I guess this makes sense.

-- 
Stan


Reply to: