The story of how the Internet learned to route itself and how it still does today.
Wait… BGP? Isn’t That Just for ISPs?
If you’re a network engineer, you’ve probably heard the term BGP thrown around a lot
sometimes with fear, sometimes with awe.
Maybe you’ve even asked yourself:
“Why does everyone say BGP is so important? Isn’t it just another routing protocol like OSPF or
EIGRP?”
Totally fair question.But here’s the thing:
BGP isn’t just a routing protocol it’s the protocol that holds the entire internet together.
And the reason BGP was invented?
It wasn’t just about getting from point A to point B.
It was about creating freedom, flexibility, and scale across a growing and chaotic global
network.
Let’s rewind the clock for a minute.
Before BGP: The Internet Had Training Wheels
Back in the 1980s, the Internet was small, experimental, and believe it or not government-
controlled.
Most traffic passed through a single “core” network managed by the U.S. government.
Routing was handled using a protocol called EGP (Exterior Gateway Protocol).
Now you might be wondering:
“Okay, so what was wrong with EGP?”
Great question.
Here’s why EGP just couldn’t keep up:
- It assumed one central routing core
- It didn’t support multiple autonomous networks
- It lacked any kind of intelligent policy control
- It couldn’t handle loop prevention or path decisions
In short: EGP was fine for a small academic network, but the internet was about to explode.
The Problem: The Internet Was Growing Up (Fast)
By the late 1980s:
- New ISPs were popping up
- Commercial networks were joining the party
- Universities and companies wanted independence
- The ARPANET was being phased out in favor of NSFNe
Each of these networks or autonomous systems (ASes) needed to route traffic on their own
terms, not just follow a central script. But with no smart, policy-based way to talk to each other… things were breaking.
That’s when two engineers stepped in.
The Birth of BGP — Written on a Napkin!
Yes it’s not just a myth. The story of BGP being drafted on a napkin is real.Here’s the quick background:
Who Created BGP?
- Kirk Lougheed from Cisco
- Yakov Rekhter from IBM
In early 1989, during a meeting of the IETF (Internet Engineering Task Force), the two engineers
realized the existing routing protocol (EGP) was no longer fit for the decentralized and rapidly
expanding Internet.They needed something:
- More scalable
- Policy-driven
- Loop-free across autonomous systems
The Famous Napkin Moment
– According to multiple sources including interviews with Lougheed and the Internet Hall of
Fame the very first draft of BGP was scribbled on a napkin during lunch at that IETF meeting.
Here’s what’s known:
The initial design concepts of BGP were literally drawn on a napkin including the basic idea
of peering, path-vector logic, and AS_PATH loop prevention.
The napkin sketch later evolved into RFC 1105, which formally introduced BGP v1 in June 1989.
Just like that, Border Gateway Protocol (BGP) was born.
- RFC 1105 was published in 1989
- First tested between Cisco and IBM routers
By 1995, BGP-4 became the standard (RFC 1771), supporting CIDR and classless routing
So… Why Was BGP So Special?
You’re probably wondering:
“Okay, but what made BGP different from RIP, EGP, or OSPF?”
Let’s break it down
- BGP Was Designed for Policy, Not Speed
Unlike OSPF (which tries to find the fastest route), BGP is policy-driven.It lets network operators say things like:
- “I’ll only accept these prefixes.”
- “Prefer ISP A unless I tell you otherwise.”
- “Don’t send traffic this way unless I prepended my AS 4 times.
- ”BGP Uses AS_PATH Instead of Metrics
BGP tracks which networks (ASes) a route passed through.
If a router sees its own AS in a path guess what? It drops the route. ✅ Loop prevented!
- 3. BGP Runs Over TCP (Port 179)
- No hello packets. No multicast.
Just a stable TCP connection between peers that keeps things clean, reliable, and controlled.
- No hello packets. No multicast.
- 4. BGP Scales Big Time
- Today’s Internet routing table has over 1 million prefixes.
BGP handles them all with ease.
- Today’s Internet routing table has over 1 million prefixes.
What Did BGP Replace?
BGP Didn’t Just Replace EGP It Replaced a Philosophy, BGP wasn’t just a technical upgrade. It was a shift in how the Internet worked:
- From centralized governance to decentralized cooperation.
- From rigid structure to flexible policy-based routing.
Suddenly, ISPs, enterprises, universities, and cloud providers could: - Peer with each other directly
- Create their own routing rules
- Connect globally on their terms
That’s the reason the Internet grew as fast as it did in the ’90s and 2000s — and why BGP is still
relevant today.
Let’s compare quickly:
| Feature | EGP (Old) | BGP (New) |
| Topology Support | Centralized | Decentralized |
| Policy Control | None | Full (filters, prefs) |
| Loop Prevention | None | Yes |
| Multiple Paths | No | Yes |
| Scalability | Poor | Internet-grade |
| Transport | Raw IP | TCP port 179 |
What If BGP Didn’t Exist?
Let’s imagine a world without BGP for a second no AS_PATH, no peering policies, no distributed Internet routing. Sounds dramatic?
because it would be.
Honestly? The Internet Would’ve Hit a Wall by the Late ’90s
No Multi-Homing = No Redundancy
Without BGP, the modern Internet as we know it would have never happened.
Here’s what we would have lost:
Want to connect to two different ISPs for failover? Without BGP, it wouldn’t be possible.
You’d be forced to:
- Manually reconfigure static routes
- Hope one provider stays up
- Pray your DNS caches hold out long enough to update
Without BGP’s route control and failover logic, the Internet wouldn’t be resilient — it would be
brittle.
Cloud Routing Would Be a Nightmare
Think about AWS, Azure, and GCP. Their hybrid connectivity Direct Connect, ExpressRoute, Cloud
Interconnect all depend on dynamic BGP routing.
Without BGP:
- You’d have to manually manage hundreds of static routes
- Cloud-to-data center paths would be fragile
- Load balancing and backup paths? Basically impossible
CDNs and Global Delivery Would Fall Apart
How does Netflix stream 4K video to you with zero buffering?
Because BGP lets them:
- Peer directly with ISPs
- Steer traffic to the closest, fastest edge server
- Adapt dynamically if a route goes down
Without BGP, content would funnel through bottlenecks, causing latency spikes, outages, and
frustration.
ISPs Would Still Be Centralized
Imagine if every ISP still had to connect to a central Internet core.That’s what the old EGP model looked like a star topology with one brain.
No flexibility. No diversity. One failure = total outage.
Without BGP, we’d still be routing like it’s 1985.
And Today?
Everything runs on BGP.
- Every tier-1 and tier-2 ISP
- Every cloud platform
- Every Internet Exchange Point (IXP)
- Every enterprise with dual WAN links
- Every data center running EVPN or MPLS VPNs
When You…
- Watch Netflix on your couch
- Check Gmail on your phone
- Spin up a server on AWS
- Access a global app from a remote village
You’re riding on BGP.
Quietly. Reliably. Every second.
So Why Should You Learn This?
Whether you’re:
- Trying to pass your CCNP or JNCIP
- Working with AWS Direct Connect or Azure ExpressRoute
- Running SD-WAN edge routing
- Or just curious about how the Internet works…
BGP is the one protocol you have to understand.
It’s not just a protocol it’s the language of the Internet.
Final Thought
What started as a napkin sketch is now the backbone of modern communication.
And as the world becomes more cloud-driven, hybrid, and borderless…
BGP isn’t going away.It’s evolving. And it’s time for you to master it from its history, to its design, to how you can use it in your
network today.