How are large organization's networks structured?
I'm specifically curious about college campuses. Do they all have a single external IP address? Or do they typically have multiple external IPs?
How are large organization's networks structured?
I'm specifically curious about college campuses. Do they all have a single external IP address? Or do they typically have multiple external IPs?
Speaking as a SysAdmin at a largeish university (20,000+ students), I have some experience with this.
In regards to one vs many IP addresses, it depends in large part on when the college got their netblock. In the mid 1980's when IP addresses were plentiful, such organizations were handed very large blocks of IP addresses; /16 blocks were common, and I think I know of at least one /9 block (or in the terms of the time, 128 Class-B networks). If the college jumped on the internet bandwagon late they got the smaller netblocks common to the .com world today. Some have even given back their large netblocks for smaller ones, though most have kept them because IP addresses are valuable.
My particular university has a /16 block, and my workstation is on a publicly routed IP address. Even though you can't get to it from outside because of firewall rules. This has hampered any IPv6 roll-out pressure, as we're feeling absolutely no pressure to move to v6 for simple IP exhaustion reasons. If we move, it'll be because we're dragged by technology or policy decisions from on high (we're a State school).
They usually have a dozen or so (or dozens for very large networks) external IPs for: various services, multiple ISP connections, redundancy, load balancing, or a multitude of other reasons. But it very much depends on the particular campus, their needs, budget, and level of expertise.
Usually they get their own IP space for a regional registry. (ARIN in North America)
With their own space they can change ISPs without renumbering.
Network Infrastructure and IP Addressing are two different questions. But since you seem to be focused on the IP addresses, look at ARIN, etc.
I work at a mid-sized university (about 10k students) and we have a /16, though we're not using very much of it.
Generally you're never going to have "one" external IP: when you sign up for sufficient bandwidth for a few hundred people, you're going to get at least a dozen addresses. If you start signing up for the kind of bandwidth a large school would require, you're going to get hundreds or thousands.
I used to work as a junior admin at a big state school and we had a godawful huge number of addresses (4 /16s a /19, a /24, and innumerable /27s and /28's). Of course, we're talking about a huge school with a huge computer science program, and multiple campi.
Generally the dorms had global ips for their routing, but for the most part the inter-dorm addresses were in the 10.0.0.0/8 block. Classrooms often had global addresses: some of the big lecture halls had their own wireless subnets (again, in the 10 block).
One of the problems with a /16 of global ips is that you have to be really careful with assigning ips, and it causes some non-trivial DHCP issues in places where you allow DHCP. In some of the older dorms we had assigned global addresses, and it caused untold headaches, especially when people started setting up their own subnets with wireless routers and crap like that. Much much easier to assign 10 based local subnets (also much easier to deploy your own WAPs rather than trying to prevent students from setting up godawful conflicting ones.)
These days I admin at a reasonably large company: we have a /16, and we hardly use it at all, preferring to have a big WAN connected on private 10. addresses. Using private address space makes a lot of things easier.