Looking for T3 service provider and Cisco Hardware Vendor

[OatMan]

Help me find the above.

800 graduate student residents and staff. Located a few blocks from Columbia University in 2 large older buildings. (Upper West Side Manhattan, NYC)

No network infrastructure of any kind for most of the residence. A small primitive LAN services the 50 or so staff. This LAN utilizes a single 10broad36 (residential cable Modem) for Internet access.

I'm putting together a proposal for a network infrastructure. I serve on the Board of Trustees for an academic not for profit international organization. The organization has recently completed a significant capital campaign and raised several million dollars. This will be used for several projects including fully refurbishing 75% of the resident rooms. This project will include significant electrical system upgrades through out the entire physical plant.

I seek to convince the Board to install a robust data inter-network. The timing of the electrical upgrades provides the opportunity to also install the infrastructure for a data network.

I am looking to provide some general numbers for the final phase of a network infrastructure project that will likely span a few years. I need to identify costs for various hardware implementations.

To do this I am looking for Vendors to provide quotes for a service contract for provisioning and support for a whole, clear T3 data line and internet service. And also Vendors to provide quotes for the network Hardware (CISCO). The specific details for the network architecture will evolve over time, but I'd like to get rough numbers to estimate hardware and support costs as best as possible. I am hopeful for a fairly simple implementation using catalyst 3500 XL switches for the access layer and majority of the 100Mbit port density. I believe a single 6500 Chassis can serve as both the core and distribution. It may be wise to use something like a 7200 router in addition serve as a gateway router.

Since I'm a budding network engineer with no experience spearheading such a project as this, I don't pretend to be able run all this on my own. But I do hope that my expertise is enough to make a compelling proposal and make a good start.

*** Any help with relevant Vendors to help me dig up some numbers will be greatly appreciated. As would any general advice.

TIA
OM

 

[spidey07]

I seek to convince the Board to install a robust data inter-network. The timing of the electrical upgrades provides the opportunity to also install the infrastructure for a data network.

And that's the way to sell it.

What it sounds like you want to do is write a Request For Proposal and submit it in writing to several network integration vendors in your area...also submit it to your local phone service providers and AT&T (so you have something to gauge the others with).

It sounds to me like you need cabling (fiber and cat5/6)
Gear (network switches, routers, firewalls)
Services - internet service and what not (I'd suggest somekind of managed internet service that also configures and maintains your internet router/firewall) You may even get a fully managed LAN if you ask for it in the RFP.

-edit- as a guesstimate you can use 600 bucks per 10/100 port - that'll cover cabling and infrastructure and 8-10K/mo for a full managed T3.

 

[Soybomb]

Might check out http://www.bandwidth.com for some quotes on your circuit

 

[cmetz]

OatMan, at the T3 level, I would suggest that you chat with (and ONLY with):

UUNet (er, Worldcom, MCI, whatever they want to call themselves this month)
Level 3 (not the best, but cheap and pretty good)
AT&T (really the old IBM network, solid)
Sprint.Net (clueful folks there)

At the T3 level, how well connected the ISP is makes a huge difference, and only a small number of players are well enough connected to deliver. Other ISPs will be cheaper, but you get what you pay for. Otherwise, you'll get a 45Mb/s connection that won't nearly deliver that in session performance to some of the other ISPs.

As far as network hardware goes, you're doing yourself a huge disservice for assuming that network equipment means Cisco. Shop around! There are competitors with better products for less money. Unless you have a problem with spending more and buying less, shop around. Even if you end up buying Cisco anyway, simply having quotes in hand from other vendors helps Cisco sales people give you more interesting discounts.

For switches, consider Extreme, Foundry, and Enterasys. For routers, consider Juniper (the M5 in particular).

A one-stop-shop integrator would be a good bet. Get a few in for proposals. Narrow the field to a small number of serious contenders, then do EXTENSIVE due dilligence on them including checking up with some reference customers. I can't tell you how many people I've seen use incompetent network integrators/consultants and have major problems. But then, I've also seen plenty of people use okay to good ones and have positive experiences.

Also, if you have bored graduate students on your network, you need to think a lot about security, because bored students means security issues.

 

[spidey07]

Good advice cmetz.

Solid, experienced, advice.

What he really needs is to have a "one stop shop" proposal from the physical layer on up and there aren't too many top-notch firms that can do that.

Project management
Cabling
Design
Security
Maintenance
Management

Wanna do some free lance consulting cmetz?
wink, wink

 

[Lord Evermore]

Go with RCN go with RCN!

I used to work for them. They're focusing on multi-megabit services like this these days. They'll set you up with a T3 and 7200 router, not sure of the price but I think maybe 10k to 15k per month was what they did when I was there but they were one-offing services then. If you're really nice, or (like Paul Allen), they might send out their head network guy to install the router and for any troubleshooting needed.

Note that I'd like for RCN to make lots of money so that my stocks become worth something again.

 

[OatMan]

Gentleman (and or ladies)

Thanks tons for the great advice. I have a few comments/questions:

1. Has anyone heard of E-Rate, or any kind of Educational discount or subsidy? I hope to leverage the Educational Institution status and or its close
affiliation with Columbia University to get a heavily discounted telecom contract. This has been suggested to me, for example Verizon does this. I don't
know if this is realistically possible.

2. I am merely using CISCO as a baseline to provide a conservative framework for estimating costs. But there can also be advantages to a complete CISCO
solution (Cisco for example will only "Officially" support their protocols on their hardware). All avenues will be ruthlessly considered [non-CISCO, used...].

3. Spidey07 - $600 per 10/100 port... is that a cost estimate for the 1st year? or per year on going?

4. The support level is a huge question. I intend to provide the range of possibilities and associated costs and pros/cons. The organization has only 2 full
time technical staff who manage both the telecom and data infrastructure. I am leaning toward suggesting hiring a 3rd full time resource to be
responsible for the network and share other responsibilities as appropriate. The reality is that they can not afford market value for such a resource, and
would use their "unique environment" to acquire such talent at a significant discount. Because of the clear conflict of interests, I do not know if I will be
able to express interest in such an opportunity myself.

5. In a nutshell I had hoped to pay for the services with a $25/month charge per residence unit. This would generate $15,000 per month and be a great
value for the cash strapped graduate student residents. I had hoped that approximately 1/3 of this would cover the telecom cost, 1/3rd the salary of
the afore mentioned network engineer and the remaining 1/3 the cost of maintenance, upkeep etc. This assumes some kind of mythological "Education
Discount". This also would NOT include the initial capitol investment (equipment, cabling, installation etc.) The financial resources for the initial capitol
investment have already been raised by a large capitol campaign which was recently completed.

 

[spidey07]

oatman,

I was talking initial capital. It is a very rough estimate to cover all infrastructure needed to deliverk network service. But also keep in mind that the capital costs for a network are generally not the biggest chuck. Support and ongoing maintenance are.

 

[Gand1]

Oatman,

Here at Kutztown University in PA we mostly us Integra for our hardware needs. With a bit of bullying we can get quite aggressive educational pricing in the %30-%35 discount range. They are a quite large company out of Allentown, PA and are a high end supplier of Nortel, Cisco, Nokia and other products.

Shoot me an e-mail if you want some names.

Integra

 

[cmetz]

spidey07, if somebody wants to write me a big check for a consulting job, I'm a capitalist, as I'm sure are plenty of other folks around here (who wants to do the trademark search for AT Consulting ). Many, MANY a consultant I've dealt with recommends that the solution to every problem is... to hire a consultant! Most consultants and integrators out there are very bad, mostly because of that wonderful property that the customers who really know the difference are generally the customers who can do it themselves to begin with. I've seen so many botched jobs that I myself tend to recommend minimizing consultants and integrators. Lesson for everyone: scrutinize consultants and integrators heavily.

I'll happily give out free advice to people as long as everyone understands that it's worth exactly what they're paying for it (nothing). I like seeing people do sane things and getting things to work. There are soooo many opportunities to shoot yourself in the foot in this business, and I'm sick of seeing people with stump legs asking for help after it's too late! Ounce of prevention.

Lord Evermore, I had a BGP routing problem once. RCN was at least part of the problem. The ISPs involved (who had links and BGP sessions to RCN) all used their contact info and I used my sources, and we never were able to find anyone with anything approaching routing clues at RCN. It was extremely frustrating and left me very sour on the IP network part of the company. I also learned in that little adventure that RCN is kinda like Verio in that it's not one network and one company, it's a bunch of networks that got squished together and one hand has even less idea what the other's doing than usual.

I find it incredibly interesting that you say they're focusing on multimeg. I thought RCN was mostly an overbuilding cable company and the rest of this stuff was other businesses they got into as diversification (Erols, Starpower, etc.)?

OatMan,

1. You're at an interesting enough price level that you should be able to do some good negotiation. You'd be doing yourself a disservice if you didn't ruthlessly play every card you could to get the vendors to all give you steep discounts. Educational, non-profit, affiliations, get several competitors in and make sure all the vendors know it, and so on. The higher the price, typically, the higher (in percentage) the margin. Which means more room to discount. At interesting price levels, MSRPs are just fiction.

Are you a tax deductable kind of educational non-profit? If so, there's some things that can be done there, too. (vendors can "donate" some stuff to you as part of the deal) Also in this case, consider whether you'd be willing to take old-generation equipment, as you might be able to find some local company who's upgrading their network infrastructure and would be happy to dispose of equipment for a write-off.

Incidentally, if you have an affiliation with Columbia, you might be able to convince them to loan you some networking folks from their IT/IS department and/or their CS department. At the very least, you should see if you can get somebody there to give you a sanity check before you execute on any plans. Assuming those folks are clueful, you should take all the free clues you can bring in.

2. It's fair to use Cisco as a reference point, but you can do a lot more for a lot less shopping around. It's true that all vendors love to finger point as soon as they find out you've got other vendors' stuff in your network. This is part of why I said find an integrator and make it their problem. That way the vendors can point fingers at each other - but you can point your finger at the integrator and make it THEIR problem, not yours (well, it's still not quite that easy, but anyway).

4. Grad students are smart, resourceful, and cheap. When I was in grad school, we were on a beer and ramen noodle kind of cost structure, and picking up a few hundred bucks a week for part time work was interesting - and common (many folks had small businesses going, consulting, web stuff, etc.). There's an opportunity there. I'd strongly suggest you do the day to day stuff yourself in house. Outsourcing it is not cost effective and puts you in a bad spot should you need to fire the company you outsourced to (now you've got nobody on-staff who understands how things work day-to-day!). Besides, you want to have some folks on site.

5. What about summer? What about people who don't pay on time? Gotta budget all those contingencies.

spidey07 is right - the buzzphrase is Total Cost of Ownership (TCO). For example, spending more on capex now to buy equipment that has good management features might make a lot of sense because it'll make it much easier to fix things later on. In the situation that you're in, it sure sounds like you should do what you can to cut down on the labor and maintenance costs as much as possible, since cash flow is the make or break variable and the telecom line itself is going to only go so low.

 

[spidey07]

you should do what you can to cut down on the labor and maintenance costs as much as possible, since cash flow is the make or break variable and the telecom line itself is going to only go so low

And at this point and density I'd be pushing for a high density or chassis based solution depending on the closets. Lets face it, having four big switches at 250 ports a piece rather than 40 24 port switches just makes common sense. No matter what the capital outlay is. With a whole mess of 24 port switches you'll spend all day fixing a problem.

But to cmetz's point, my large network projects are projected over 5 years (3 years straight depreciation, plus upgrades/expansion for the full life of 5 years). Once you start doing the math the initial capital is pretty small.

 

[OatMan]

Great advice cmetz.

I spent 5 years as an enterprise platform system integration consultant. Mostly at large pharmaceuticals. I worked with all kinds, so I know the double edged sword consultants can be.

I am trying to identify different plans for every level of "do it yourself". From a completely managed solution for the entire inter-network, to a complete in house scenario with nothing more than a managed incoming line. There is a stiff learning curve, but the time line is generous.

spidey07

And at this point and density I'd be pushing for a high density or chassis based solution depending on the closets. Lets face it, having four big switches at 250 ports a piece rather than 40 24 port switches just makes common sense. No matter what the capital outlay is. With a whole mess of 24 port switches you'll spend all day fixing a problem.

For now this is all at a high level, so not that much delving into the plumbing, so to speak, but since you mention it, this is what I was thinking. The physical considerations are going to be difficult. The residence house is a sprawling neo classical monster build in the '20s. It will be difficult to find space for wiring closets and such. The "resourceful" graduate students also present a great security challenge. There fore I figure to use one "gateway" router to do the heavy lifting and boarder the outside world. One router to partition the Staff infrastructure. And then one or two routers for the residents. They are in two buildings, so it may make sense to use two routers instead of just one. I figure that only the gateway router will need to flex most of the muscle, so the others don't need to be anything great.

I wonder about how to best concentrate the port density. If I centralize then there will be a huge number of cable lengths running long distances. If I decentralize I want to avoid exactly what you describe. I hope to have One large chassis for the smaller (resident) building. One large chassis for the staff network in the second building. And then perhaps two more chasses for the remainder of the larger builder (remainder of residents) one on the east side one on the west.

I guess this is rhetorical since there isn't anything for anyone to go on... But this is really getting ahead of myself first is merely winning a commitment for the concept.

 

[OatMan]

actually come to think of it, the gateway probably wouldn't be doing most of the work...

 

[OatMan]

UPDATE!

Well, I gave my "Exploration of Concept" presentation to the Board committee. It went very well. I believe they were quite impressed, I know the President and Directors were. If nothing else I have convinced them that they need a long-term technology plan and will be working to help develop it. We'll see where this goes... I think I outlined the dangers of Wireless a little too thoroughly; they're looking under their beds for hackers!

Thanks a ton to everyone for all the great input. It was INVALUABLE!!!

My hope is to start seeking RFPs from Vendors this fall.