5

I'm faced with a bit of a dilemma with regards to stretching a 100base-T network from building to building.

The site is a racetrack (cars) with a control tower, housing race control, admin and a small air conditioned server room. About 200m away, across the assembly area, is a newly built Hospitality Suite.

Someone, in their infinite wisdom, decided that both buildings needed separate telephone lines, and ADSL connections, and gave no thought to interconnecting the two separate copper networks.

We recently developed and implemented a live race results service, delivered by locally hosted webpage, and the powers that be have decided they would like to project said live results onto some of the big TVs in the hospitality suite.

We tried a wifi link, but there's an awful lot of radio traffic floating about a race circuit, and the reliability was very poor (even with directional Yagis fitted and pointed at each other).

We learned recently that there is a duct running from building to building underground, with a fair bit of room to spare, since all we need is a connection between two existing copper networks (one of the ADSL lines is going, that's not an issue), should we be looking at Fibre switches? Or can we just run some Cat5e to a media converter on both ends?

Complete fibre newbie here, help appreciated.

Thanks!

2 Answers 2

4

You're saying the right things, in my opinion. For building-to-building connectivity fiber can't be beat. It will be the last cable you ever need to run between the buildings (provided you run enough). Strngly consider running multiple strands for redundancy and future expansion. I generally recommend running no fewer than 6 strands.

Once you get this up and running you can think about pushing other traffic over the fiber-- eliminating excess phone lines and using VoIP, etc.

100Base-TX over multi-mode fiber (called 100Base-FX) is a pretty straightforward application, and I think you'll have good luck with it. You should look at costs for gigabit Ethernet hardware, too, though. If you can afford to bring the fibers up at gigabit speeds now I would. Even if you in-building networks are 100 megabit now eventually you'll replace switches and end up with gigabit speeds.

Installing and terminating the fiber isn't really something you want to do yourself if you haven't been trained to do the work. You'll spend less money and time, and get a reliable infrastructure, if you have a contractor perform the installation.

You certainly can use Ethernet switches with fiber optic ports in them, but media converters are a perfectly viable route, too. I've used a lot of the little Allied Telesys FS-202 media converters over the years with great success. (I know of one pair that have been running in a site for almost 10 years.) They're inexpensive (under $200.00 each) and work well.

On the other hand, if you have switches that already have GBIC, SFP, or dedicated multi-mode fiber ports you might have some cost savings using them.

I tend to like a media converter over an GBIC or SFP because I retain the option to swap out the switch w/o affecting the fiber-to-copper conversion. OTOH, using standalone media converters means keeping a spare around in case of media converter falure.

4
  • 2
    Good advice, even if I would strongly recommend going for SFP's/GBIC's/XFP's and not media converters. Cheap mediacoverters can mangle your datalayer, creating hard to debug situations. It's also one more thing to administrate, power and so on.
    – pehrs
    Mar 6, 2010 at 15:03
  • Thanks, to both answer posts. The organisation is voluntary, so our equipment has to be justified with more than just "this will work well". I'm sure they'll like the possiblity of the switches with SFPs, modularity and ease of replacement make their eyes light up. As regards to the ducting, can fibre be pulled the same way as any other delicate cable, with respect to minimum bend radius etc? We already have someone who pulled the cables already in the duct, but I don't know if he has any experience with fibre. How difficult is terminating the fibre, is there a lot of room for error?
    – Bradfirj9
    Mar 7, 2010 at 12:32
  • The fiber does not have any special installation demands compared to copper. Being non-conductive it's actually considerably easier to pull them, as you don't need to worry about induction. If you are reusing an old duct make sure the person installing knows what he is doing, as it's easy to destroy both cables when pulling a new one in. For the termination of the fiber you need to splice on a patch panel. This is done with a fusion splicer. They are expensive and requires training, let a contractor do that for you. He can also measure the fiber and give you a protocol.
    – pehrs
    Mar 8, 2010 at 8:29
  • Terminating the fiber is the tricky part. It requires some specialized tools and for your task, some expensive consumables as well. Some good tutorials arcelect.com/fibercable.htm and lanshack.com/fiber-optic-tutorial-cable.aspx and terminating the fiber: hyperline.com/slide/optic
    – chris
    Mar 8, 2010 at 13:15
4

A pair of moderate price gig switches with SFP would cost short money. The optics would also be super-cheap.

Have a company pull a conduit between here and there and have them pull a bundle of 12 single-mode fibers through that conduit.

The optics for are slightly more expensive for the single-mode but you'll never ever have to upgrade it. If you decide on multimode, make sure you use 50 micron fiber instead of 62.5, and make sure your patch cables are also 50 micron. Multimode over that distance may not ever "support" 10gig / 40gig / 100gig, which is why I suggest using singlemode fibers. It would probably work, but given the small cost differences between SM and MM, it is probably better to just do it right from the start.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .