New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
What would be the benefit using FO over Wire Copper on 1 Gig
Dear Folks,
It seems I can't see nor find one reference (article) that could clearly state the benefit of Fiber Optic Cable, except that it could serve long distance better, over Cat6 Cable. What would you say? Is there any differences? Other benefit, mostly, as FO would be an expensive investment than Wire Copper?
Comments
I believe it's mostly to do with the quality of the connection, less signal degradation, which leads to a larger possible bandwidth throughput.
If you are running this between buildings over the air, or from the top floor downwards, when there's a thunderstorm optics will safer since static electricity can't accumulate in the wires to burn out your endpoint NICs/switches (not to mention if there's an actual lightning strike nearby).
Copper cable
@century1stop you can do 10G on copper (10GBASE-T) with Cat 6a+
uh? There are DAC cables with QSFP+ up to 40Gbit.
Multimode fiber (OM3/OM4) is now generally cheaper per meter than Cat7.
Fiber NICs cost more though, especially on Gbit. 10G is around the same but fiber is cheaper used (eg. Myricom cards).
100M on Gbit, yea. OM3/OM4 also only does 300M though.
A 3R interconnect has 0% loss but costs, a copper patch HAS considerable loss.
folks.... 1 gig please. I'd have open another thread with 10 gig
Copper is just fine for 1Gbps as long the distance won't be a bottleneck. Copper patches / port setups when connecting to a transit/peering pop is usually cheaper as well.
Most people @ DC would go for LR to have the possibility to upgrade to 10GE over their existing line. 10GE over fiber is usually cheaper than a 10G copper port.
Stability wise no difference if you ask me. If you don't need 10G I would go for 1G copper. No chances that a optical module goes black for example. 1G copper should run rock solid.
What exactly is "long distance" here?
That was my point. IN rack scenarios, I don't think 'long distance' would be necessarily needed, except if you want to cross connect between racks (from floor 1 to 16 )
How do you feel in using both for those who have already been implementing both scenarios? Basically in terms of scientific, the speed of light would be much faster (100x times or more I believe) compare to electricity transfer). But I'm just curious in "production" state.
Copper ethernet @ 1Gbps using Cat 5E / 6 has a 100 meter (~330ft) limitation. Still quite a distance though :-).
1Gbps fiber, multi-mode (SX) has with OM2 fiber a 550 meter limitation
1Gbps fiber, single-mode (LX) has usually a limitation of 5KM
And with 'special' equipment (ZX) you can even reach 70KM over an optical 1G link. We use for example 10G-ZR (80KM) for our 10Gbps DC-to-DC connectivity :-)
Using single mode fiber is pretty much a standard. Pretty much the whole internet is connected with single mode fiber. Multi mode is something you usually only see within internal networks.
So it all depends on your requirements. If you exceed 100 meter of cable length, you'll be fine with using copper. If you're exceeding this distance, 1Gbps SX is probably the most cheapest option to light up 1Gbps over a distance up to 550 meters.
On top of my previous message. Optical links actually have "slightly" more latency, as there is an additional layer where optical is being converted into electrical signals. You don't have that with copper links.
Something you don't really notice in real life situations though, but it's funny to mention though :-)
There is really no functional difference between fiber and copper as long as they are running within their respective physical/environmental specifications.
For the most part, you use fiber when the distance, speed or environmental requirements of copper are prohibitive.
I don't think "cheapest" is very meaningful in the context of the year 2016.
isn't it more efficient to have fiber to fiber connectivity than fiber to copper?
What do you mean by "efficient"? The Ethernet protocol is the same on either medium.
In the context of physical facilities, you could say fiber is more "efficient" when you consider xWDM. But that has nothing to do with this particular situation.
@Microlinux, I agree. Both using the same technology to "convert", conversion from Analog to Digital. Electrical conversion is faster? Interesting in reading all those comments above, yet, there are still no articles / references pertaining the benefit of using FO over Copper other than distance benefit
There are facts that scientist need 10year++ to discover a robust material above 1G using copper, because of the interference' shielding and other obstacles in delivering "pure data" using high-grade/quality signaling on copper. I think they will stop in doing that in couple years ahead as the cost vs benefit doesn't seem to match.
What would I think of them in the future is that they will develop an FO cable that would as harder (good) as a copper (the PROs of using copper will be adapted to the FO), and would drop copper's further developments. So we'll be talking about giant leap in the IT standards, omg.
It's basically isolated. The only real use for fiber inside the home is either 1) for a fast NAS, or 2) for really low latency between computers (assuming you have the fiber cards).
That's funny, considering 1G fiber does NOT have any speed or latency advantages over 1G copper.
try this article http://blog.blackbox.com/technology/2015/04/8-advantages-to-choosing-fiber-over-copper-cable/
Only in specific DWDM/CWDM wavelengths though, i wish you good luck sourcing a 12??nm CWDM 10G-ZR optic....
1G was even extended more, there are 120km optics on single-mode (even single cable BiDi) in 1310/1550, they work fine but i would be very worried about a 120km non-pumped link....
Further notes:
1G copper converter SFP modules work... for some time. They tend to fail randomly.
80/120km optics generate a not to be forgotten amount of heat also due to non-perfect (read: near null) air flow in most switches
Optic coding is a scam, nearly anything will eat any optic (with log complains) but warranty might be limited (EU -> lolnope). The fancy coding devices and blank optics sold make sense even for just a few 10 often.
10G Switches for cheap? -> Force10/Dell, but only L2 and IPv6 support partly questionable (read: software). Cheap is relative, be warned.
BSD (FreeBSD, OpenBSD, Pfsense etc.) tends to have issues with some used refurb cards with a Terminator 3(?) chipset, you'll see these easily by looking for a PCIe 8 lane 1.0 interface (much lanes due to age of standard).
Forget Infiniband unless you can source cables for cheap (China?) as only the cards are cheap (might be bit old status, but my 2x 20G cards did cost like 35EUR each and each 1m cable like 25EUR...). It's also a nightmare to use natively (RDMA, exception is GLusterFS which works ok actually despite being a nightmare itself) and if you like IPoIB... yea support for that is limited (very) and the protocol overhead (Infiniband is very low level) costs you 25%+. Cheap to link 2 servers together at "ethernet" speeds of realistic 2x 15Gbit (cannot bond IPoIB) though, with just around 100EUR total.
Ebay is your friend, 10G HW is everywhere since the large enterprises upgrade to 40/100G. Exception sadly switches as they tend to eg. end up in QA/staging envs which just had Gbit before and now get 10G, while production is upgraded to latest.
DR it through before posting. Sorry