(Not sure if this is an appropriate group for this question; sorry if it isn't.)
We currently have a 100Base Ethernet network and are moving to a new location so taking advantage of the situation and upgrading our network, possibly to multimode fiber. I've spent the past couple days scouring the net about fiber and have a few questions for which I cannot find answers (note that we're not going all the way to the desktop with fiber, but hopefully at least to the wall plates so we're ready in a few years when 1gig and 10gig network cards become standard):
- Ethernet uses switch(es) to connect all the workstations. Most of the "fiber switches" I see only have connections for a small handful of fiber lines and a bunch of ethernet ports. So, how does one physically plug a few dozen fiber connections (with workstations at the other end) together and to the main switch? I think this is my big question as I'm obviously just not getting something. I read that we won't need switches (or closets) on different floors due to the longer cable length, so how do all the fiber lines from each floor "merge" and feed to a single port on the main network switch? Or do I need to buy a bunch of fiber switches to be able to handle upwards of 64 lines?
- I see a bunch of stuff about 10/100Base fiber. Why would someone use fiber if they're only getting 100Mb/s?
- I've been all over the place on the web getting bits and pieces of info, but can anyone recommend a site (or even a book) about how to actually put together a fiber LAN? What parts are needed? etc. (No, I'm not doing it myself, but want to have a clear understanding of how it works so I can estimate costs, etc).
I'm no expert but the above paragraph strikes me as silly. If you take fibre to the wallplate and no further, you'll have to convert it to copper there an then. Much easier to do that in the wiring closet.
GigE is doable already with copper between desktop and the wiring closet, and with careful planning 10GigE might be too. Whether you should want 10GigE to the desktop when there is now no actual need for GigE is something different entirely. What is your planning scale? Five years? Ten? Copper will likely be around for a while yet.
If you are about to commission laying cable, you could read up a bit about how the experts would do it, over in comp.dcom.cabling.
That would be because one typically does the last ``hop'' with copper, and the long haul stuff that needs more bandwidth with fibre. If you really want fibre to the workstation, you get switches that have all fibre connections. And probably a lot of spare patchcables for the last meter or so to the desktop.
Remeber that if you have 100 workstations that somehow manage to fill up their 10GigE connections, you have very little use for 10GigE backbones. You'll need to come up with a backbone that can carry, oh, a terabit.
Apart from the longer reach, fibre is very useful in industrial settings and for spanning between buildings because it isn't conductive, avoiding interference, ground loops, that sort of thing. Lightning strikes in the vicinity can induce enough potential in conductive cables to destroy attached equipment or even the cables themselves.
you buy an expensive switch that supports lots of fibre ports or cards - often that means a chassis of some sort.
some like a Cisco 4500 ot 6500 would work.
Note fibre ports seem to suck more power than copper, cost more, need expensive optics modules, and dont as dense cards.
I think this is my big
you can get 48p Gige SFP cards for cisco. But with a csico pluggable fibre multimode SFP running at $500 list, i doubt you will fully populate a chassis full without a good reason.....
it goes a long way. lots of old systems need long distances or non electrical cables - think of links around a power switching substation.
GigE on multimode is limited to 250 or 500m (depending the fibre type).
The only actual fiber to the end computer I've actually discussed with someone was when they installed DEC mini-computers into an aluminum smelter up in the Pacific Northwest in the 80s. It usually just isn't done unless you have severe magnetic/electrical/security concerns. I'm sure a lot of government workstations use fiber to the desktop but that's a market where budgets don't mater like in the "real" world.
As to the smelter site, the fellow told me a game they'd play was how many paper clips could you stack up end to end. The fields were so strong they could get 8 if careful. After that the arc as they followed the curve of the field would tip over the column. :)
Use fiber to connect your switches and copper to the desktop.
I work for a government. We have spending constraints tighter than the majority of private businesses. For example:
- Our departmental budget was fixed more than 10 years ago, and has been chopped a couple of times since then. The relevant elected officials have made it clear to us that our budget will not be increasing any time in the foreseeable future (that is, they have taken a decision to no longer fund through that mechanism)
- The number of continuing employees the department is permitted was capped and may only be changed by a specific vote by the legislatures. The number is over the department as a whole, so if a continuing employee is hired in one region, the number of continuing positions available in other regions is decreased.
- There is a limit to the length of time an employee may be hired for term positions (5 years I think it is at the moment.) It is not a simple a matter as not making long-term commitments to individuals: we have to actively get rid of them.
- Salary increases (including cost-of-living increases) for all continuing employees come out of the fixed budget; heating costs also come out of the fixed budget. Thus, the fixed budget "buys" less and less over time, effectively requiring that we get rid of continuing employees in order to cover costs. As noted above, we cannot just convert those employees to indefinitely renewed terms: we cannot keep individuals on term for more than a few years.
- We are not permitted to advertise in any way.
- If we do manage to find outside contracts or "fee for service" then any revenue we earn that way which has not been spent by the end of the fiscal year will be taken back by the government.
- All desktop computers, Windows servers up to 4-way, Linux servers up
8-way, and telecommunications equipment (including switches and routers) must be chosen from a "standing offer" list of pre-packaged systems, said list being updated approximately once a year, unless it can be proven to the satisfaction of the auditors that none of the equipment on the list was able to meet the reasonable requirements of the situation. For example, you could not say that you have a VOIP requirement and choose a non-listed switch on that basis, unless you had a specific plan in place to implement VOIP: a -potential- future need is not a "reasonable requirement" compared to a -current- requirement to use the standing offer list when possible. If you are Cisco shop with decades of experience in Cisco boxes and Cisco did not happen to make the standing offer list for the kind of telecommunications equipment you need, then you would not be allowed to buy the Cisco unless you could *prove* that the retraining of your staff would cost more than would be saved by buying from the list. And as best I recall, only "hard costs" such as training courses and travel costs for the courses are counted as retraining costs, with "soft costs" excluded, such as migraines and "opportunity costs" from trying to get the other box to do what is easy with the accustomed hardware.
The IT infrastructure situation around here is greatly oriented towards "do more with less"; it is hard to replace equipment that is not actively failing. Productivity costs and opportunity costs are not taken into account unless they can be proven -- because if they cannot be proven (rather than "best professional recommendation in line with standard business practices") then there is a risk that someone will put up a legal challenge and say "That was just your opinion". Risk management and trend analysis don't win legal challenges: the legal cases are loaded and we lose them if there was the -possibility- that the provable costs associated with the challenger -might- have been lower. Who knows, we
*might* have had a flash of inspiration and suddenly completely grokked a system voted "Worst User Interface" three years running.
Saying that government offices are not operating in the "real" world is perhaps true in some senses, but it is a mistake to think that government offices always have money to spare. I know that we sure don't!
Jeff, I have a very good testimonial for fiber to the dektop
formatting link
use of fiber can really be a cost benefit depending on your application. Cut & paste the url above and if that seems to be of intrest let me know. Also there are various grades of multimode fiber, if you do go fiber make sure that it is nothing less than 50 micron and perferabby lasor grade.
There are areas of federal government spending where price doesn't mater much at all. And many of these area intersect with 0 tolerance for EMF leakage. This is the kind of installation I was alluding to. DIA, NSA, etc... And I know folks who've worked there and for some tasks you just spend money until it's done. And the next day you spend 2 hours justifying your paper clip usage.
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.