BGP - The Real-World Guide (Not Boring)

📡 BGP - The Real-World Guide (Not Boring)

BGP (Border Gateway Protocol) is how routers across 
different networks (Autonomous Systems - AS) talk 
to each other to exchange route information.

🌐 BGP Types - eBGP vs iBGP

Type	Description	Preferred When
eBGP	Between DIFFERENT ASes	Connecting to ISP, cloud, external networks
iBGP	Inside the SAME AS	Sharing BGP routes within your own routers or data centers

🤝 BGP Neighbours (Peers)

BGP needs neighbours to share routing info. You can use physical IPs or loopbacks.

✅ Use loopbacks + “update-source” for stability ✅ Use “ebgp-multihop” if not directly connected

## 🔁 eBGP - External BGP

✅ Preferred when:
- Connecting to another AS (e.g. ISP, cloud provider)
- Advertising your public IPs
- Receiving external internet routes

🧠 Notes:
- Default TTL = 1 (direct connect)
- Use `ebgp-multihop` if not directly connected
- You CAN use `network` command

---

## 🔄 iBGP - Internal BGP

✅ Preferred when:
- Sharing routes between routers in the same AS
- Data center or enterprise BGP backbone
- Redistributing eBGP-learned routes internally

🧠 Notes:
- iBGP needs full-mesh or Route Reflectors
- Next-hop is NOT changed by default → IGP must reach it
- You CAN use `network` command here too

---

## 🔃 iBGP Full-Mesh vs Route Reflector

| Design          | Preferred When                                      |
|------------------|----------------------------------------------------|
| Full-Mesh        | Small networks (3–5 routers)                       |
| Route Reflector  | Larger networks (5+ routers), ISPs, DC cores       |

🧠 RR allows non-full mesh by letting one router reflect routes to others.

---

## 🌍 Loopback vs Physical Interface

| Interface Used   | Preferred When                                          |
|------------------|--------------------------------------------------------|
| Loopback         | You want stable BGP peering over multiple paths        |
| Physical         | Simpler setups with direct cables                      |

Use: neighbor X.X.X.X update-source loopback0

🔢 BGP network Command

✅ You can use network in both eBGP and iBGP

🧠 How it works:

• Tells BGP: “If this route exists in my routing table, advertise it”
• Must match exact prefix
• Good for manual control

✅ Preferred When:

• You want to control what routes are advertised
• Advertising static or IGP-known prefixes

---

🔄 Redistribute into BGP

✅ You CAN redistribute static or IGP routes into BGP
⚠️ Use with caution — not preferred without filtering

❌ Not Preferred When:

• You want stability and clean route control
• You don’t have route-maps or filters

✅ Preferred When:

• You have many dynamic routes to advertise
• You’re okay managing filters (prefix-lists, route-maps)

---

🧰 Route Filtering Tools

Tool	Purpose
Prefix-List	Allow/block specific IP prefixes
Route-Map	Modify BGP attributes (LP, MED, tags, etc)
Distribute-List	Basic filtering (rarely used with BGP)
Communities	Tag routes for easier policy control

---

⚖️ Controlling Traffic – BGP Attributes

Attribute	Used For	Preferred When
Local Preference	Outbound traffic (higher = preferred)	Choose best exit from your AS
AS Path	Inbound traffic (shorter = preferred)	Influence how others come to you
MED	Suggest best inbound route (lower = better)	You have multiple links to same neighbour
Weight (Cisco)	Local to the router only	Tie-breaker, not shared with others

---

📦 Use Cases – What’s Preferred?

Use Case	Preferred Setup
Advertise public prefix	eBGP with network
Share loopbacks internally	iBGP with network
Connect to 2 ISPs	Dual eBGP + AS path prepending
Share eBGP routes internally	iBGP with route reflector or full-mesh
Prefer one ISP for outbound	Set higher Local Preference
De-prioritize inbound route	AS Path Prepending
Remote peering (not directly connected)	Use loopback + ebgp-multihop

---

🚨 What’s NOT Preferred

❌ iBGP without full mesh or route reflector
❌ Redistribute BGP into IGP (can cause loops)
❌ Advertising too many routes without filters
❌ Not ensuring next-hop reachability in iBGP

---

🧯 BGP vs IGP – When to Use Each?

Protocol	Preferred For
OSPF / EIGRP / IS-IS (IGP)	Internal routing (fast convergence)
BGP	External routing, policy control, WAN scale

Use IGP for underlay (reachability), 
Use BGP for overlay (control, flexibility)

---

✅ Quick Recap Table

Goal	Preferred Method
Connect to ISP	eBGP
Advertise internal services	iBGP + network statement
Internal reachability	IGP (OSPF, EIGRP, etc.)
Multi-router internal setup	iBGP with Route Reflector
Stable peering	Loopbacks + update-source
Load balancing out	Local Preference
Load balancing in	AS Path Prepending or MED
Avoid iBGP full mesh	Route Reflector
Avoid route leaks	Prefix-lists + route-maps

---

🔥 Final Tip: 
Always filter what you advertise. 
The Internet doesn't like surprises.

---

🙌 Connect With Me