I'm curious about how people approach network connections between separate buildings, especially when reliability and electrical isolation are important.
Copper Ethernet can be simple and inexpensive for shorter runs, but once equipment is located in different buildings, factors like grounding differences, lightning exposure, and surge protection can become more important.
This is why I'm interested in when people decide that fiber is the better option.
For those who have worked on multi-building or outdoor network deployments, what usually influences your decision the most?
Electrical isolation between buildings
Protection from lightning and surge issues
Distance limitations
Higher bandwidth requirements
Existing conduit or cabling infrastructure
Cost and installation complexity
One thing I'm also wondering about is the practical setup on the remote side. If fiber is used between buildings, do you normally place a switch with SFP ports at each location, or do you prefer using media converters for smaller installations?
I can see fiber being a clear advantage for larger networks, but I'm interested in hearing where people draw the line for smaller setups. At what point do you consider fiber worth the extra installation effort compared with copper?
I'd appreciate hearing about real-world deployments and any lessons learned from installations where copper caused problems or where fiber turned out to be unnecessary.
When you're thirsty, is it better to have a glass of clean water or a glass of dirty water?
No...
but really...
it doesn't make any sense...
Unless there's some physical obstacle, (that would prevent ALSO the copper cable from being installed),
just install fiber.
Any other consideration about what to choose is ridiculous, if you can bring a cable (of any type).
Then you can discuss the type of fiber, media converter, SFP, or whatever, but the rest is completely pointless...
Yes, it is the better choice, fiber is always the choice for inter-building links.
Anything thats longer than interrack (DAC), or inter-premise (usually rj45 or even fiber sometimes depending on factors where and how long), fiber is always the choice.
Each building or even each part of a larger building has its own grounding pole buried into the ground. Between two grounding poles there may be voltage differences between these grounding points. If you connect them a large current may flow.
Standard Ethernet uses transformers for isolation and removel of dc components. This is only a functional isolation, not a safety isolation. So you may use Ethernet only for components connected to the same grounding point. You must use fiber for outside connections. This is not a choice. Depending on the local power system and laws there might be regulations for this.
Fibre wherever possible, and native on the switches/routers (no media "bastardizers" - just one more thing to fail). I also run SFP+ interfaces on servers and other high speed interconnects that can support it.
The main question is more "single or multi mode"
(Heck, I even run a 10G optical backbone in my house between routers and switches, as well as core server connections. I don't see anywhere it would be a bad choice.)
I see 10G on copper mainly as a solution in search of a problem . . . .
DAC's are always better in every way than fiber at short distances especially in the same rack, unless theres a long distance inbetween, or its inter-building/inter-site run, SM fiber is also cheap and better than MM everywhere and it really doesnt matter which one you use unless we are talking about multi 100 gigabit links.
Singlemode fiber transcievers are more expensive than MM fiber transcievers, thats why people go for it at home and for shorter runs.
This is true, long ago I worked at some place where buildings had separate groundings but they were joined together at the corners (so you could walk from one to another building). There was one network cabling for all buildings in this cluster. The young, female secretaries wore sometimes short skirts during summer and were always electrocuted when touching with their legs the metal-housing of their PCs.
Temporary fix was using unshilded LAN-cabels to the PCs.
To the TO, use always fiber! It doesnt matter if its a big or smol setup. Fiber, or if not possible, a Radio-Link. I would not use copper (beside the limit of <100m per link).
As I always say, if there's room for Ethernet, there's also room for 24 fibers...
Oh, I haven't written this yet,
but I'm assuming that if you run a cable with fiber inside, I mean at least 12 fiber cable...
Not exactly a single fiber inside, or a single pair...
If it's less than 100 meters between the buildings and you don't need more than 1 GbE just run UTP. It'll be the quickest and cheapest option to install.
I assume you're going to have some kind of watertight duct between the buildings regardless of whether you use UTP or fiber, and in that case ordinary unshielded indoor Cat6 will work just fine as long as it stays dry and isn't exposed to UV.
The idea that buried UTP automatically has problems with grounding differences, lightning exposure, and surge protection is one of those myths spread by incompetent amateurs.
If you're going with fiber and haven't done it before, and don't have someone who can help you choose the right kind of fiber, optics, equipment, installation, and configuration, I'd get a company that knows this stuff to help you out. Otherwise, there's a pretty good chance you'll end up getting something completely wrong.
Whether you do it yourself or get outside help, expect it to be considerably more expensive than UTP.
If you're looking for 10 GbE, then fiber is the natural choice. See point two regarding installation and equipment selection.
Please write your opinion. Have you ever prepared BOM and cost estimate?
What decided what solution was better?
It is next topic with an artificially created open question that looks like pumping users to get "stories" to publish them or needed for a summer assigment. Moreover it could be a "loooong question" to feed an LLM with data from real users.
If there would be no sensible answer the topic probably would be closed.
Someone asks a perfectly reasonable technical question and your first reaction is to accuse them of farming stories, doing a summer assignment, or feeding an LLM?
If you think the question is artificial or pointless, just don't answer it. There's no need to derail the thread with this kind of speculation.
Why not?
Do you think that there would be any answer from the OP?
Is the OP involved in a topic in any way except that question that is not a question?
Because you're making assumptions about the OP's motives instead of discussing the technical question.
Maybe the OP comes back, maybe not. That doesn't really matter. If the question itself is relevant and others can contribute useful experience, then the thread has value regardless.
If you don't think it's worth answering, just move on. There's no need to turn every thread into an investigation of whether it's LLM-generated or whether the OP is "real enough". All you're doing is derailing and cluttering the threads with this.
It's not a question if OP is real enough. Can you comment how these topics ended?
What would be next? What PCs do you use? What .... you use as "it is known that ...blah, blah".
Well to be fair, look at their activity, https://forum.mikrotik.com/u/rosiecharles/activity, it looks like broad questions that someone would ask a llm to answer for them.
To me it does look like an attempt on feeding ai data through official mikrotik forum. Because people often who use llm to get their answers tell it to look at OEM forums or reddit for answers and best practices.
So i do kinda see @BartoszP 's point.
Or setting up own WWW page with some "technical" information it is nice to be able to make references to an old forum posts that "prove" truths on the new site.
Those topics may or may not have gone anywhere but that doesn't justify derailing new threads by speculating about whether they're LLM-generated, some kind of assignment or whether the OP is "real enough".
If a question is too generic or low-value then just don't answer it or explain briefly why it's too vague to be useful. Turning the thread into a discussion about the AI or OP's motives just creates more noise than the original question.
You miss the point too. I just asked for an involvment and pointed that BOM and cost estimation tell you the truth "why fiber is better than copper". Real involvent. Simple.
Especially that:
If there would be none then "it could be assumed" that it was a bot. Isn't it? Other topics confirm that.
