Calling all network admins Part 2

[Silvertwinkie]

Ok, second part:

You have an existing CAT 5 cable plant that represents about 85-90% of your entire deployed plant. The workstations you currently buy are all capable and ready to deal with gigabit to the desktop, and your network gear is or soon will be ready to deliver gigabit to the desktop, if only your cable plant could support it. You've done testing and found that for the most part your CAT 5 deployment will not carrry the signal to gigabit speeds. So here is the question, knowing you have to replace the cable plant within your buildings:

1) Do you save money and simply replace to CAT5e so that the cable plant can fully support gig to the desktop?

--OR--

2) Would you spend slightly more money and put in a CAT6a solution that would have the certified potential of carring signals up to 10 gig?

In researching this, I have found the price of the cable to be a bit more expensive, but I've also found that the biggest cost to all this is the labor.

So what would you do?

[overlander63]

Eric, if you have the budget, go with the CAT6a. Network speeds will only increase as time goes on, and you will save money over time by not having to upgrade twice. You will also look like a far-seeing genius to the kids that will be using it.

[lewster]

Twink,

A good friend of mine who does high-end specialty wiring in VERY expensive homes once told me (with his thick German accent) 'viring is cheep, de labor is not! Don't vorry about de vire costs and always use de best!'

Hope this helps in your decision.

[herrgirdner]

Good advice, Lew
regards

[smily]

The minumum standard for the Air Force is Cat 6 whereas the Navy utilizes Cat 5e. The cost of Cat 6 is considerbaly higher in many aspects. Not only do you need the high dollar cabling but the end components are VERY expensive. Installation is not so much different but testing criteria is much more stringent hence costing more.
Installing cabling more than once, (read upgrade) is always going to be additional cost, obviously. Why pay more than once for increased throughput demands?

One could speculate that a fiber optic solution is the only "meets all upgrades" approach. Fiber to the desktop is getting more and more common everyday. A number of years ago some Engineers had the foresight and realized that copper soulutions are riddled with high overhead and subject to upgrades approximatley every ten years (if not less). These Engineers chose a fiber optic to the desktop solution years ago. The impact was the expensive NIC cards but the infrastructure has not required any change. One other drawback to Fiber Optic NICs is that many users like to interface with third party networks which are typically copper oriented. I have engineered and implemented fiber to the desktop solutions for many government agencies and the one thing that has been consistant is that network throughput requirements continually increase, whereas the properties of a fiber optic infrastructure meets the increase every time.

Go with fiber.....

[Silvertwinkie]

Yea, I thought about the fiber bit, but the reality is in our business, we work with over 2200 children in an elementary school setting. Fiber to the desktop isn't practical at this point, even though it's far superior to any copper plant I could install. It may the case where the ancillary cost of fiber (patch, NIC, etc) make it more reasonable, but the numbers are too close at this point, and in our case buying machines like Intel based iMacs, adding NICs isn't a simple endevor.

So, I love the fiber idea, actually have about 20 classrooms with fiber in them that was pulled many years ago...have never used it....yet.

For now, the choice is CAT5e (1gbs standard) or CAT6A (10gbs standard). CAT6 won't do 10gbs....from what I understand it's a hybrid CAT5e.

[herrgirdner]

I do agree that the B/W of a fiber installation is outstanding, and if left undisturbed, will outlast and endure over copper. However, and ahem.
With 2200 active children exploring the F/O connectors, I could imagine a full-time employee running around just redoing them. When I was terminating the stuff, we used exoxy to cement the fiber to the little middle rod, and then cleaved it, and then polished it to a fare thee well. I understand that technology has obsoleted that procedure, thank God.
I read one comment where a test bed was recommended. Not a bad idea.
regards

[thecatsandi]

Go with the highest bandwidth copper available. The wire is cheap. The labor is expensive. Wired a whole building with cat 5e about 10 years ago.

[Westfalia]

I don't see the advantage to running a 10gbps capable link to your clients if the receiving client can't process 10gbps worth of data. As it is, 1gbps connections are hardly utilized to their fullest potential in a end client situation, usually topping off at 200-400mbps in real world situations.

Where a 10gbps link is going to be useful in is backbone networks or for high throughput server links where the hardware had been built to handle this kind of link.

So, I would run Cat-5e and max at 1gpbs for your clients. You could always drop a line or two of Cat-6 in each location.

[Silvertwinkie]

I don't see the need today either, but then again, I also didn't think in less than 5 years we'd be at 1gbs to the desktop, nor did I think I'd see some workstations start to appear that were diskless on the horizon where the weakest link, the hard drive, might not be an issue. I guess part of my current thinking is that you put in stuff like fiber and CAT6a, not so much for what you can think of, but for what you can't and if the costs are not too far from what say a CAT5e might be.......

Thanks for all the comment and insights, I appreciate everyone's thoughts!

[Westfalia]

Your points about 'future proofing' are completely valid. However, for the next five years, I don't see links higher than 1 gbps to the end-client becoming the norm for a number of reasons, most namely due to hard disk speed. For the most part, disk seek and R/W speeds have only slightly improved over yester-years PATA-100. There will be no way to store and process 10gpbs of data on the receiving end even if you can jam that much down the pipe. Also, take into account that most of the cheaper computer use an on-board NIC (MacMini included). Unlike a dedicated NIC card, The faster link chews up more processor power than the slower link. It is doubtful (to me at least) that the budget computer will move to dedicated processor NICs in the future due to cost and I don't think that any budget computer made in the next five years will have to processing power to move 10gbps. Then again, smarter people than me have been wrong about these thing before!

I guess the ultimate deciding factor is cost: what is the price differential between Cat5e and Cat6 installed for this job?

[Silvertwinkie]

Ok, so let's play the bit per byte game. 1gbs is about 112megabytes. Currently there are disks in some configs that do some serious throughput. We've been assuming that all traffic would be disk based, but what if the machines evenutally in 5 years start not to have hard drives, or that the multimedia streams are not directly connected to the hard drive but the web or other internal streaming resources. It is true that the current 1gbs is the limit, but say 10 years from now, which if done right is hopefully the average cable plant life that isn't the case.

Trust me, I like the next person wants to spend as little as possible and get the biggest bang for the buck...but is CAT5e the way, cause even if data gets to 5gbs and diskless hard drives happen, and the sun still comes up tomrrow, is it really worth spending the labor costs to put 5e in?

I'm don't totally disagree, and clearly can't tell what beyond 5 years will bring, but I have a hunch it may be more than the current 1gbs to the desktop.

[SRoffe]

Wow, I usually read these threads at work when talking with fellow broadcast engineers. I'm enjoying this.

With 2200 elementry school student, and the demands of multimedia in the future, perhaps CAT6a would give you beyond 5 years life out of your infrastructure. Labor cost and cabling cost will go up in 5 years, that is for sure. So, if you can get more life out of your cabling, you might be ahead of the curve in the long run. ( 10 years maybe )

I have to agree with you though, who knows what the standard will be 5 years from now. It may be with fiber, and cost of fiber NICs will be affordable like the NICs of today.

Sam

[GetOutDoors]

Why not CAT6e? It's stranded, like CAT5e, but has the higher through-put.

[Silvertwinkie]

It was my understanding that the Cat 6e solution is the same as the Cat 6a. Some folks call it e while other a, the specs seem identical (625mhz, shielded, though there is an unshielded Cat 6a too).

There was one version of 10gbs cable, not sure what it was called, but that cable could also do 10gbs, but it could not do it up to 100m, the current locations of our network cabinets to the farthest workstations.....

[herrgirdner]

Some questions that drifted through... Are you soliciting bids from a number of BICSI qualified contractors? Do you have a trusted consultant in the wings? Assuming that you are deciding on the intrabuilding cabling for an ethernet distribution system and that the cables would be homerun from switches to the enduser work stations, would a cabling plant upgrade imply that the various switches and servers also be upgraded? There may be some implicit but (as yet) hidden costs in some of the more advanced technological solutions.
Interesting thread.
regards

[Silvertwinkie]

All good questions.

Yes, I am asking for proposals and am using a certified network designer specializing in cable plant deployments.

The issue of electronics has been addressed for current standards (10ge backbones, 1gbs desktop), so the cable plant that goes in should be able to support at first a 1gbs transmission to the clients from the access switches, but as the access switches age and need replacement, the cable plant should be able to support more than the current 1gbs standards. As I wave my crystal ball looking beyond year 5, I get the sense that there is a real possibility in years 6-7 (after deployment) that 1gbs to the desktop will be what 100mbs to the desktop is today and that higher capacities may be not only possible, but probable.

In my particular case right now, most of my existing plant can't even support 1gbs to the desktop, and in Illinois, for major projects in public schools, when you go out to the community for additional funding beyond normal funds collected from property taxes (due to tax caps), you are required to spend the $$$ within 3 years of approval. These fundings called referendums happen about one or two times in 10 years, if that, with no promises that even if you float one, the local community would approve it, so while we have these funds now is why we are looking at this as we just addressed the issue of the electronics which were end of life, end of support. The weakest link in my chain right now is the cable plant. Currently, 75% of my workstations and laptops support 1gbs as does my switchgear.

[GetOutDoors]

Originally Posted by Silvertwinkie

It was my understanding that the Cat 6e solution is the same as the Cat 6a. Some folks call it e while other a, the specs seem identical (625mhz, shielded, though there is an unshielded CAT6a too).

Originally Posted by Silvertwinkie

If my understanding is correct, CAT6 defines the through-put standard and the letter after designates the type of wire core. The letter e is suppose to specify small stranded wire cores, which is used for applications that need a more flexible core, like in a TT, where road vibration can take it's toll on a more rigid core. I used CAT6e in my Safari for my private (non-wireless) network, and used the extra for sat. controls.

[GetOutDoors]

Ok, just for future references to this thread . . .

A quote from a website that I would name, but I don't know if I'll allowed to do that here, so . .

"The general difference between Category 5e and Category 6 is in the transmission performance and extension of the available bandwidth from 100 MHz for Category 5e to 200 MHz for Category 6. This includes better insertion loss, near end crosstalk (NEXT), return loss and equal level far end crosstalk (ELFEXT). These improvements provide a higher signal-to-noise ratio, allowing higher reliability for current applications and higher data rates for future applications."

Further, I am told that manufacturers have the liberty of "tweaking" the way they make wire, so long as they maintain the minimum standards of the CATEGORY. That said, you may get slightly better performance of a verses e, depending on the manufacturer, but your network speed is also dependant on the hardware you use (patch panel, routers, cable connectors, etc.) for speed. So you can have a cable that is better, but still not be able to capitalize on it.

Calvin