What are the different deployments used in practice to connect network

[makeupideaz]

EIGRP is an interior gateway protocol used for routing purpose in enterprise networks to connect different branch offices, headquarter and their private datacenters over a WAN network. What are the different network deployments that are practiced commonly with EIGRP in connected networks of enterprises? Below are the three topologies which I could think of -

1. PE-CE deployment over MPLS L3 VPN cloud.
2. Over MPLS L2 VPN cloud - where each branch router would have many EIGRP neighbors on a particular intf just like connected to a big virtual switch.
3. DMVPN

Are there any other deployment strategies? Or correct me if my understanding is not right.

[paul driver]

Hello
TBH of all those three you mentioned , I have yet to come across any large network deploying EIGRP due its earlier proprietary with cisco plus most networks are not just running all cisco hardware.

[MHM Cisco World]

for DMVPN is excellent choose to use EIGRP, the EIGRP support summary at each router and that we so need in DMVPN. 

for MPLS Core EIGRP not recommend because it not support MPLS-TE 
for MPLS PE-CE you can use it but as I know it not so recommend, the reason is if Core use OSPF then using CE-PE ospf make it easy for ISP admin to run two process in one router than use two protocol in one router, also some feature like shim-link can use simply.
if you use CE-PE BGP the admin can do some filter path control that can not done if use EIGRP CE-PE

[Joseph W. Doherty]

In the (usually pretty large) Enterprises I've worked in, they all used OSPFv2 for internal (IPv4) routing.  They also used OSPF (even when all devices were Cisco's, in theory, allows much easier migration/incorporation of other vendor equipment - also useful when bargaining with Cisco over equipment pricing [where there is much wheeling-and-dealing, when you're buying millions of dollars of equipment]) for routing between locations when using WAN technologies like p2p private links, frame-relay, ATM, and p2p VPN or DMVPN.  When dealing with PE-CE over MPLS VPN clouds, generally we would use BGP, between sites (because, at least at that time, they often didn't support running OSPF/EIGRP across their cloud).  In the last company I worked in, a very large Enterprise, they also used BGP across their (internal) WAN cloud.  (Also, when "adding" IPv6 to the Enterprise, they decided to use IS-IS.  [I jumped up and down saying, nooo, use OSPFv3, but as this company also is a very large ISP, using IS-IS for both IPv4 and IPv6 inside their ISP part of the network, they decided all would be good.  {Last I heard, about a year after I left that company, they decided to replace IS-IS for IPv6, within the Enterprise, with OSPFv3.}])

As for WAN/MAN topologies that appear and work much like Ethernet (i.e. your "big virtual switch"), most, I believe, just use whatever IGP they've been using.

PS:

In you're wondering why I was against using IS-IS for IPv6, vs. OSPFv3, it's not so much one is "better" than the other, it was mainly due to, in the Enterprise, most network engineers are familiar with EIGRP/OSPF rather than IS-IS (which is more often found within ISP networks), and, even more importantly, much "Enterprise" or SMB devices will support OSPF, but not IS-IS.  (In fact, when we were going to replace our 3750s [original series, through -Xs], with 3850s, the 3850 [at the time] did not support IS-IS.  So, we went with another vendor's small L3 switches that did support IS-IS.  [Oh joy, let's learn IPv6, a totally different IGP and another vendor's much different configuration syntax all at the same time.  Perhaps good for one's personal resume, but it does seem to lead to many "teething" problems, and interesting different vendor equipment issues.  [Like, discovery, new vendor's equipment, on triple speed Ethernet ports, didn't support 10/half - because no one uses that anymore - well except we had some very old, special proprietary equipment, which couldn't be replaced with anything better/newer, and only used 10/half.])

[MHM Cisco World]

@Joseph W. Doherty  BGP in MPLS Core is use but for LDP You need routing protocol like OSPF. 
BGP use only for VPNv4 not for connect router inside Core. 

[Joseph W. Doherty]

As, within the very large (about 100,000 employees, 5,000 "enterprise" network devices, hundreds of sites, even a few overseas) Enterprise I worked, the Enterprise network, and the ISP network, were managed almost in the way that we (Enterprise) were just another customer of our own company's ISP network (which, on the network equipment capacity side, made our Enterprise look small).  I.e. we were not directly involved in their network, nor they in ours.  However, even though we were different "networking" teams, we were still in the same overall management structure, and did exchange information that was not done with "real" external customers.  Also, personally, I started with this company in the operations support group, which principally supported the ISP portion of our networks.

Forgoing said, it was over a decade ago, I worked in operations, and over four years ago, I worked in their Enterprise network.  I.e. my memory can be off.

On the ISP side, we were running MPLS, natively, and, yes, LDP was being used.  I recall (?) for IPv4 IGP, being used within the MPLS core, as being IS-IS, but, perhaps it was OSPFv2, and perhaps it wasn't used totally within the MPLS core.  (Again, so long ago, I'm unsure of all the particulars.)  I do recall, they did use IS-IS for IPv6, but not all the particulars how they used it.  (In fact, since we were all one huge network group, the primary reason they decided to use IS-IS, within the Enterprise network, for IPv6, is than ALL our network engineers, would have one common IGP, for IPv6.  [BTW, ISP and Enterprise, were the two major demarcations of our networks, but there were lots of different networking engineering groups, supporting various parts of our networks.])