You've got Office applications, heavy development environments, all the applications you need. They are in a virtual environment, safely locked away at the corporate HQ datacenter. That way they can keep them patched and secure, and you can use them from anywhere you've got Internet access.
This could make virtual desktop deployments easier. Many IT organizations want to deploy VDI, but have to deal with mobile workers and their laptops. It's hard to prove an ROI when mobile users have a full laptop AND VDI licenses.
Now, a mobile user can have one mobile device (phone) and use the nearly ubiquitous 4G and Wifi connectivity to get to remote desktops back at HQ's datacenter. No doubling up mobile devices with a tablet.
Not if you want to support it. A small library could easily need
-an onsite Domain Controller (for AD failover)
-Switchports for those 8 computers and 10 PoE phones and 5 APs to provide wireless coverage
-WAN acceleration. 2 or 3 Citrix machines? Maybe the library app uses SQL to communicate with a server somewhere else?
There you go. A UCSX module, a WAAS module, and an Etherswitch module. All on one support contract, too. I'm not saying they need those services, but if enough sites use them, it makes sense to deploy that capability everywhere. That small rural library may very well have wireless, VDI, and other capabilities in the next few years.
That is all well within reason. And I would disagree that a consumer device would work. No IT department in their right mind would want to support 1000 consumer boxes. They fail too often, don't support advanced services (see above), and the wireless coverage from a single omnidirectional antenna can barely cover a 1000 sq ft house, much less a municipal library of who-knows-how-big. These 1000 sites are being managed centrally. RBAC with AD integration and a real LMS would be crucial to handle config changes, OS upgrades, etc.
(I work for Cisco)
That's a lot of expansion modules - those could be switches, WAN accelerators, ESX servers... all sorts of things.
There's cost savings in management when you put services in the router instead of separate boxes. Plus, then you don't buy separate boxes, too.
(I work for Cisco, if that matters)
One possibility lies in the "expansion modules" discussed.
A 3945 is, indeed, a potent router capable of handling WAN connections over 100Mb. That's way more than you'll probably find in rural WV, today.
If this was bid to an RFP of some sort, then it gets muddy. Who wrote the RFP? Why did they choose those capabilities?
However, a 3945 Integrated Services Router (Generation 2) also support 4 SRE modules. These routers _could_ have integrated switch modules. No one will ever have to go out there to console into the switch - because it's a card in the router. There could be WAAS or UCSX modules, which provide WAN acceleration and ESXi hypervisor capabilities. There could be VOIP SRST capabilities built in for future (or current) voice redundancy. Again, this seems expensive, but generally shows an improvement in management down the road. One place to manage all the equipment in the library can be a significant improvement.
That router _could_ be replacing a WAN accelerator, a key system, a firewall, a switch, and a small VM server. Or maybe it's gratuitously oversized. The article doesn't include enough information to make that decision.
Lastly, 1000 T-1 cards added $1M to the cost. Well, yeah. That means each card cost $1k.
(I work for Cisco, if that matters)
The math doesn't work out - the majority don't live in dense areas. There are just lots and lots of people in the not-dense areas due to their large size. Check the lists of most dense cities in the US. See how quickly the density falls off:
Also, if you add up the population of the top 100 cities in the US (which includes many not-dense areas) you'll only have 20% of the population.
I'd like to point out that "most" Americans don't live in dense urban areas. In fact, very few do.
First, many cities are very spread out, particularly in the south. This lack of density makes mass transit more difficult. So just because Houston, Dallas, Birmingham, and Atlanta have large populations doesn't mean that they have dense populations - they cover large areas. Houston (the fourth largest city behind NYC, LA, and Chicago) has half the population of LA and 20% more area. So, it only has 40% the population density. Phoenix, San Antonio, San Diego, San Jose and Dallas all have population densities less than half of LA.
Here's a list of all the incorporated places with population density over 10,000 per sq mi. Notice how all of these areas are parts of just a handful of large cities.
Second, a minority of Americans live in big cities.
If you add up the populations of the top 100 cities (dense or not) in the country you'll only end up with 20% of the population. Only 9 cities have over 1M inhabitants. The 100th largest city is the bustling metropolis of Spokane, WA with just over 208k people.
Here is the list of cities by population:
(Video discusses population density as part of an Electoral College discussion. Start at 3:30 for the population density discussion)
So, please understand that your hip urban life in LA has little in common with where "most" Americans live. A minority of Americans live in cities of any real size, and of those only a fraction live in highly dense areas.
Lastly, compare those population densities with the density of European cities. Paris has twice the density of NYC:
Also - go watch the rest of CGPGrey's videos. They are awesome.
It is much easier to suggest solutions when you know nothing about the problem.