So, I started out just trying to build a simple network.

Nothing complicated —
some Linux nodes, a couple CE routers, a small core with a few PEs and Ps —
just enough to get basic end-to-end IP connectivity.

My goal was straightforward:
✅ Bring up the nodes.
✅ Get them talking across the network.
✅ Basic pings from Node1 to Node2, maybe Node1_2 to Node2.
That’s it.

At least, that’s what I thought was going to happen.

🛠️ The Lab Setup

The topology started simple:

• Node1 ↔ CE1 ↔ PE1 ↔ P1 ↔ P2 ↔ PE2 ↔ CE2 ↔ Node2
• Node1_2 ↔ CE1_2 ↔ PE1

Each link had its own /30 subnet.
Each router had a loopback address.
I set up OSPF across the core.
Enabled MPLS forwarding.
Built LDP neighbor sessions between all core routers.

Before even getting to my original "tests,"
I already had full:
✅ OSPF neighbor adjacency
✅ Loopback-to-loopback IGP reachability
✅ LDP neighbors formed across the backbone
✅ MPLS label-switched paths between all routers

🎯 First Test: Pings

I spun up the nodes.
Set IP addresses and default routes on the Linux devices.
Built static routes on the CEs pointing upstream to the PEs.

Fired off a ping from Node1 to Node2...
And immediately saw:

Ping from Node2 to Node1?
Same thing.

But when I pinged from a Node to its CE, or from the CE to its PE?
✅ Success.

When I tried pinging further?
❌ Nothing.

🧠 The Realization

At first, I thought something was broken.
Maybe a static route was missing.
Maybe OSPF wasn’t redistributing something.

But the more I looked, the more obvious it became:

✅ Every router in the core knew about the other routers' loopbacks.
✅ LDP sessions were up.
✅ Labels were installed.
✅ CEs could reach their directly connected PEs.

The problem wasn't the core.

The "problem" was design.

🚦 Why Nodes Couldn't Reach Each Other

In this setup:

• The Linux nodes are connected to CEs.
• CEs have static default routes pointing to PEs.
• PEs have static routes pointing toward their local CE-facing interfaces.
• The MPLS backbone is pure transport only — it doesn’t know about customer routes.

And that’s the whole point.

This lab didn't break.
It worked exactly how it's supposed to work at this stage.

In a real SP design, generally speaking:

• The core (P routers) only runs the provider’s internal IGP (like OSPF or IS-IS) and MPLS label switching.
• The core does not carry customer routes. Ever.
• PE routers are responsible for keeping track of customer prefixes — but they isolate those routes inside VRFs (Virtual Routing and Forwarding tables).
• Customer traffic moves across the backbone using labels, not IP lookups.
• iBGP (VPNv4 address family) is used between PE routers to exchange customer routes across the MPLS core.
• End-to-end customer reachability only exists once BGP/MPLS VPN services are layered on top of the MPLS transport — not before.

Without VRFs and VPNv4 BGP between PEs,
there's no way for PE1 to know "how" to reach Node2 sitting behind PE2.

📋 What Was Actually Working

✅ The Linux nodes could ping their directly connected CE router.
✅ CEs could ping the PE router they were attached to.
✅ Nodes could even ping the egress interface of the PE they connect to.

And that's good.

It proves:

• Static routing between the CEs and PEs was working.
• The PEs were reachable from their connected customers.
• The PEs were isolating the customer traffic exactly the way a real SP backbone should — no leaking into the core, no shortcuts.

The fact that the Nodes could ping up to the PE egress —
but couldn’t reach remote Nodes through the core —
was exactly the confirmation that the backbone was doing its job.

✅ Local reachability = Good.
✅ Global customer reachability = Not allowed yet (by design).

🚀 What's Next

Now that the transport core is up and running:

• I’ll build VRFs on PE1 and PE2.
• Bring up iBGP between PE1 and PE2 over their loopbacks.
• Exchange VPNv4 routes.
• Then finally deliver full Node1 ↔ Node2 reachability through the MPLS cloud.

That would be Phase 2.

But, like it always goes —
Now I’ve got multiple Phase 2s stacked up waiting to be done.

Every time I finish one thing, I end up starting something different —
another lab idea, another way to test something new, another rabbit hole to dive into.

Maybe at some point I’ll go back through and stitch a few of them together into one big Phase 2.
Maybe.

Either way — this foundation is ready for whatever comes next.

📑 Topology and Configs

As always, I am using EVE-NG and I have provided the base configs that I used.

🛠️ Here’s the Logic I Used at the Time:

✅ MPLS labels ride across the core.
✅ CEs just do basic IP, no dynamic routing.
✅ Full control plane separation between customer and provider.

🌎 Loopback Assignments (Used for OSPF Router-ID and LDP)

PE1 (Juniper vMX)

set system host-name PE1
set interfaces lo0 unit 0 family inet address 1.1.1.1/32
set interfaces ge-0/0/0 unit 0 family inet address 10.0.0.1/30
set interfaces ge-0/0/1 unit 0 family inet address 10.0.5.1/30
set interfaces ge-0/0/2 unit 0 family inet address 10.0.10.1/30
set interfaces ge-0/0/2 unit 0 family mpls

set protocols ospf area 0.0.0.0 interface lo0.0
set protocols ospf area 0.0.0.0 interface ge-0/0/2.0

set protocols mpls interface ge-0/0/2.0
set protocols mpls interface lo0.0

set protocols ldp interface ge-0/0/2.0
set protocols ldp interface lo0.0


set routing-options static route 10.0.3.0/30 next-hop 10.0.0.2
set routing-options static route 10.0.6.0/30 next-hop 10.0.5.2

P1 (Cisco IOS)

hostname P1
!
interface Loopback0
 ip address 1.1.1.2 255.255.255.255
!
interface GigabitEthernet0/0
 description To PE1
 ip address 10.0.10.2 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/1
 description To P2
 ip address 10.0.11.1 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/2
 description To P4
 ip address 10.0.13.1 255.255.255.252
 mpls ip
 no shut
!
router ospf 1
 router-id 1.1.1.2
 network 10.0.10.0 0.0.0.3 area 0
 network 10.0.11.0 0.0.0.3 area 0
 network 10.0.13.0 0.0.0.3 area 0
 network 1.1.1.2 0.0.0.0 area 0
!
mpls ldp router-id Loopback0 force
mpls label protocol ldp
mpls ip

P2 (Cisco IOS)

hostname P2
!
interface Loopback0
 ip address 1.1.1.3 255.255.255.255
!
interface GigabitEthernet0/0
 description To P1
 ip address 10.0.11.2 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/1
 description To P3
 ip address 10.0.15.2 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/2
 description To PE2
 ip address 10.0.12.1 255.255.255.252
 mpls ip
 no shut
!
router ospf 1
 router-id 1.1.1.3
 network 10.0.11.0 0.0.0.3 area 0
 network 10.0.15.0 0.0.0.3 area 0
 network 10.0.12.0 0.0.0.3 area 0
 network 1.1.1.3 0.0.0.0 area 0
!
mpls ldp router-id Loopback0 force
mpls label protocol ldp
mpls ip

P3 (Cisco IOS)

hostname P3
!
interface Loopback0
 ip address 1.1.1.4 255.255.255.255
!
interface GigabitEthernet0/0
 description To P2
 ip address 10.0.15.1 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/1
 description To P4
 ip address 10.0.14.2 255.255.255.252
 mpls ip
 no shut
!
router ospf 1
 router-id 1.1.1.4
 network 10.0.15.0 0.0.0.3 area 0
 network 10.0.14.0 0.0.0.3 area 0
 network 1.1.1.4 0.0.0.0 area 0
!
mpls ldp router-id Loopback0 force
mpls label protocol ldp
mpls ip

P4 (Cisco IOS)

hostname P4
!
interface Loopback0
 ip address 1.1.1.5 255.255.255.255
!
interface GigabitEthernet0/0
 description To P1
 ip address 10.0.13.2 255.255.255.252
 mpls ip
 no shut
!
interface GigabitEthernet0/1
 description To P3
 ip address 10.0.14.1 255.255.255.252
 mpls ip
 no shut
!
router ospf 1
 router-id 1.1.1.5
 network 10.0.13.0 0.0.0.3 area 0
 network 10.0.14.0 0.0.0.3 area 0
 network 1.1.1.5 0.0.0.0 area 0
!
mpls ldp router-id Loopback0 force
mpls label protocol ldp
mpls ip

PE2 (Juniper vMX)

set system host-name PE2
set interfaces lo0 unit 0 family inet address 1.1.1.6/32
set interfaces ge-0/0/0 unit 0 family inet address 10.0.12.2/30
set interfaces ge-0/0/0 unit 0 family mpls
set interfaces ge-0/0/1 unit 0 family inet address 10.0.4.1/30

set protocols ospf area 0.0.0.0 interface lo0.0
set protocols ospf area 0.0.0.0 interface ge-0/0/0.0

set protocols mpls interface ge-0/0/0.0
set protocols mpls interface lo0.0

set protocols ldp interface ge-0/0/0.0
set protocols ldp interface lo0.0

set routing-options static route 10.0.2.0/30 next-hop 10.0.4.2

📦 CE Routers and Linux Hosts (Simple)

CE1, CE1_2, CE2:

• Static route to PE
• Default route from PE to CE

Node1, Node1_2, Node2:

• Static IP
• Default gateway to CE

CE1 (Juniper vMX)

set system host-name CE1
!
set interfaces ge-0/0/0 description "To PE1"
set interfaces ge-0/0/0 unit 0 family inet address 10.0.0.2/30
!
set interfaces ge-0/0/1 description "To Node1"
set interfaces ge-0/0/1 unit 0 family inet address 10.0.3.1/30
!
set routing-options static route 0.0.0.0/0 next-hop 10.0.0.1

CE1_2 (IOS-XR)

hostname CE1_2
!
interface GigabitEthernet0/0/0/0
 description To PE1
 ipv4 address 10.0.5.2 255.255.255.252
 no shutdown
!
interface GigabitEthernet0/0/0/1
 description To Node1_2
 ipv4 address 10.0.6.1 255.255.255.252
 no shutdown
!
router static
 address-family ipv4 unicast
  0.0.0.0/0 10.0.5.1
!
commit

CE2 (Juniper vMX)

set system host-name CE2
!
set interfaces ge-0/0/0 description "To PE2"
set interfaces ge-0/0/0 unit 0 family inet address 10.0.4.2/30
!
set interfaces ge-0/0/1 description "To Node2"
set interfaces ge-0/0/1 unit 0 family inet address 10.0.2.1/30
!
set routing-options static route 0.0.0.0/0 next-hop 10.0.4.1

Node1 (Linux)

ip addr flush dev eth0
ip addr add 10.0.3.2/30 dev eth0
ip link set eth0 up
ip route add default via 10.0.3.1

Node1_2 (Linux)

ip addr flush dev eth0
ip addr add 10.0.6.2/30 dev eth0
ip link set eth0 up
ip route add default via 10.0.6.1

Node2 (Linux)

ip addr flush dev eth0
ip addr add 10.0.2.2/30 dev eth0
ip link set eth0 up
ip route add default via 10.0.2.1

🎯 Connectivity Verification Checklist

Before I realized why the Nodes couldn’t reach each other,
I ran through some basic connectivity checks to make sure the core transport was actually working.

Here’s what I verified:

I'm not posting 20 different images of ping results. So, these images are only parts of the verification and in no particular order

✅ Node to CE Pings

✅ CE to PE Pings

✅ PE to P Core Pings (loopbacks)

✅ P to P Core Pings

❌ Node to Node Pings

✅ OSPF Neighbors

✅ LDP Neighbors

✅ MPLS Forwarding Table

🧱 Wrap-up

OK — I don't want to bore you with any more verifications and checks.

Everything that needs to work, works.
Everything that shouldn’t work (yet), doesn’t — and that’s the point.

What started as a quick lab to test "something" (I forgot now), turned into a solid foundation for a full MPLS-enabled Service Provider core.

Along the way, I ran into a few dead-ends, found clarity at 2AM,
and reminded myself that training under pressure (or exhaustion) just makes you sharper —
or forces you to go to sleep.💤

Now I’ve got a stack of Phase 2 labs I want to do.
Like I said earlier – maybe I’ll go back through and cleanly roll a few of them together in one big Phase 2.

Or maybe I’ll get distracted and start something else.
We’ll see.

But this core is done.

Thanks for reading —
catch you in the next one.

— Bryan ☮️ OUT!