The first time I saw BGP running in production, I’d been a network admin for three years and didn’t know what I was looking at. We were cutting over to a new ISP, and the senior engineer kept saying things like “I’ll prepend the AS path before the maintenance window.” I nodded. I didn’t know what an AS path was, why you’d prepend it, or who I was supposed to be peering with.
That gap catches a lot of CCNA holders off guard. You finish the cert thinking you understand routing because you can configure OSPF in your sleep, then walk into your first real enterprise environment and find a whole layer of routing logic that wasn’t on the exam at all.
Border Gateway Protocol is the routing protocol that makes the internet work. Every public IP address you’ve ever reached was reachable because BGP told one network how to get to another. It’s not on the CCNA 200-301 anymore. It used to be, briefly, on the 200-125, but Cisco pulled it when they consolidated the exam in 2020.
This article walks through what BGP actually does, the ways it differs from the protocols you studied for the CCNA, and why most working network admins end up having to learn it sooner or later.
What Is BGP and What Does It Do?
BGP, short for Border Gateway Protocol, is the routing protocol used to exchange reachability information between autonomous systems on the public internet. Version 4, defined in RFC 4271, has been the standard since 2006. Every router on the public internet that talks to another network is either running BGP or sits behind something that is. There is no alternative at internet scale.
The job BGP solves is policy routing. It does not look for the fastest link or the lowest latency. It carries information about which networks are reachable through which neighbors, along with attributes that let administrators say things like “prefer this ISP over that one” or “do not advertise our customer prefixes to that peer.” Decisions get made based on business agreements as much as technical metrics.
This is also why BGP feels different to work with. OSPF and EIGRP both ask the same general question, which is what’s the cheapest path through the network I control. BGP asks a different one entirely, which is which routes do I want to accept from each neighbor and which do I want to send back. When you sit down to configure it, you spend less time thinking about cost values and more time thinking about route filtering and neighbor policy.
Why Isn’t BGP on the CCNA 200-301?
Cisco removed BGP from the CCNA when they collapsed the old two-exam track (ICND1 and ICND2) into the single 200-301 in February 2020. The previous version, the 200-125, included a small BGP objective: “Configure and verify single-homed branch connectivity using eBGP IPv4 (limited to peering and route advertisement using Network command only).” That was the extent of it. Basic peering, one neighbor, the network command. Nothing about path attributes, route maps, or iBGP.
When the 200-301 launched, that objective got cut. The new blueprint kept OSPFv2 single-area as the only dynamic routing protocol in the exam, alongside static routing and first hop redundancy protocols. BGP, EIGRP, and IS-IS all moved up to CCNP territory, where they’re covered in the CCNP Enterprise track.
The reasoning Cisco gave was that the CCNA needed to broaden into automation, programmability, and security fundamentals, and something had to give. Dynamic routing got narrowed to OSPF. Whether that was the right call depends on who you ask. Working network engineers tend to think BGP should at least be mentioned given how common it is in real environments, while a lot of instructors point out that the blueprint is already overstuffed without it.
How Is BGP Different From OSPF?
They solve different problems at different scales. OSPF is an interior gateway protocol, or IGP, designed to find the shortest path inside one administrative domain. BGP is an exterior gateway protocol, or EGP, designed to exchange routes between separate administrative domains.
The differences play out across almost every dimension of how the protocols behave:
| Property | OSPF | BGP |
|---|---|---|
| Protocol type | Link-state IGP | Path-vector EGP |
| Designed for | Inside one organization | Between organizations |
| Transport | IP protocol 89, direct | TCP port 179 |
| Neighbor discovery | Multicast hellos | Manually configured |
| Best path metric | Cost (based on bandwidth) | Path attributes and policy |
| Convergence speed | Seconds | Slower, often minutes at scale |
| Typical use | Inside the data center, campus, WAN | Edge to ISP, multi-site, internet |
The convergence number is the one that surprises people. After studying OSPF and watching it reconverge in a few seconds, the BGP convergence behavior feels broken at first. It’s not. It’s intentional. The protocol is designed to be slow and careful because at internet scale, fast reconvergence is how you accidentally blackhole half a continent.
What Is an Autonomous System?
An autonomous system, or AS, is a collection of IP networks operated under a single administrative policy. Your ISP is an AS. AWS is an AS. So is Google, Cloudflare, your cable provider, and any large enterprise that runs BGP at the edge. Each AS gets a number, called an ASN, assigned by a regional internet registry like ARIN, RIPE, or APNIC.
ASNs used to be 16-bit, giving a total range of 1 to 65535. That ran out around 2011, so the standard expanded to 32-bit. The current range goes up to 4,294,967,295. Private ASNs, used inside organizations that don’t peer publicly, sit in 64512 to 65534 for the old 16-bit range and 4200000000 to 4294967294 for 32-bit.
When you peer with another network using BGP, the first thing you exchange is your AS number. Your router and theirs compare ASNs to figure out whether the session is eBGP or iBGP, and that distinction changes how attributes get processed.
What’s the Difference Between eBGP and iBGP?
eBGP runs between routers in different autonomous systems. iBGP runs between routers in the same autonomous system. Same protocol, different behavior depending on whether you’re talking to a peer outside your AS or a peer inside it.
The most important practical difference is what BGP does with routes it learns. By default, routes learned over eBGP are advertised to all other BGP peers. Routes learned over iBGP are not re-advertised to other iBGP peers. That last part exists to prevent routing loops within an AS, but it has a side effect: if you want all your internal routers to know about external routes, you either need a full mesh of iBGP sessions (every router peering with every other one) or you use route reflectors or confederations to scale that out.
The other difference that matters is the TTL on the BGP session itself. eBGP defaults to a TTL of 1, meaning the two routers have to be directly connected at Layer 3. iBGP runs over your internal network, so the TTL is set high enough to cross multiple hops. If you ever see eBGP sessions failing to come up between two routers you know can ping each other, the TTL is usually where to look first.
How Does BGP Choose the Best Path?
BGP uses a long sequence of tiebreakers to pick a single best path when it has multiple options for the same destination. Cisco’s implementation walks through them in this order:
- Prefer the path with the highest weight (Cisco-specific, local to the router)
- Prefer the path with the highest LOCAL_PREF (shared across the AS)
- Prefer paths originated by the local router
- Prefer the path with the shortest AS_PATH
- Prefer the path with the lowest origin code (IGP over EGP over Incomplete)
- Prefer the path with the lowest MED (multi-exit discriminator)
- Prefer eBGP paths over iBGP paths
- Prefer the path with the lowest IGP metric to the next hop
- Prefer the oldest received path
- Prefer the path from the neighbor with the lowest router ID
- Prefer the path from the neighbor with the lowest peer IP
In practice, most decisions get made at the top of that list. Weight and LOCAL_PREF are the two knobs that get adjusted most often. Weight lets one router make a local decision about which path to prefer. LOCAL_PREF lets an AS as a whole agree on which exit point to use. AS_PATH length is where things get interesting, because prepending your own ASN multiple times into the path you advertise makes that path look worse to other networks, which is the trick that senior engineer was talking about during my ISP migration.
If this looks like a lot to keep straight, that’s because it is. Most working network admins do not memorize all eleven steps. They know weight, LOCAL_PREF, AS_PATH length, and MED, and they look the rest up when they have to.
When Will You Actually Use BGP at Work?
You’ll touch BGP if any of the following describe your environment:
- Your company has more than one internet circuit and needs to influence which one is used for outbound or inbound traffic
- Your company runs its own AS number and advertises its own IP space publicly
- You connect to cloud providers over Direct Connect, ExpressRoute, or Cloud Interconnect, all of which use BGP for the underlay
- You work for an ISP, a managed service provider, or any organization with multiple sites that peer with each other or with carriers
- You run SD-WAN, which uses BGP under the hood for a lot of overlay route exchange
The cloud connectivity case is the one most enterprise admins hit first. A typical AWS Direct Connect setup has you running eBGP between your edge router and AWS. The configuration is small. The concepts behind it are not. If you’ve never seen BGP before and the first time you encounter it is during a cloud migration project with a deadline, that’s a bad time to be learning.
How Do I Start Learning BGP?
Start with eBGP between two routers in a lab. You can do this in Cisco Modeling Labs, GNS3, EVE-NG, or any platform that supports IOS images. Building a home lab is the same exercise as for OSPF, just with BGP loaded.
A reasonable progression looks like this. Get a basic eBGP session up between two routers in different ASNs. Advertise a single prefix in each direction. Verify the routes show up with show ip bgp. Then add a second connection between the same routers and watch BGP pick one as the best path. Change LOCAL_PREF on one side, watch the decision flip. Add AS_PATH prepending and see how that affects what the other side prefers.
Once eBGP feels comfortable, set up an iBGP session between two routers in the same AS and notice what doesn’t get re-advertised. Add a third iBGP router and see what happens when you don’t have a full mesh. That’s the lesson that makes route reflectors make sense.
The Cisco Press CCNP Enterprise study guides have solid BGP chapters if you want a structured path. Cisco’s BGP configuration guide is dense but authoritative. INE and CBT Nuggets both have BGP tracks aimed at CCNP candidates. None of these are required reading for the CCNA itself. They become required reading the moment you take a job at a company that runs BGP.
BGP FAQ
Is BGP on the current CCNA exam?
No. BGP was removed when Cisco launched the 200-301 in February 2020. The current blueprint covers static routing, OSPFv2 single-area, and first hop redundancy protocols as the routing topics. BGP is covered on the CCNP Enterprise track instead.
What port does BGP use?
BGP runs over TCP port 179. The TCP session has to come up before any BGP messages get exchanged, which is why a firewall rule blocking 179 will silently break peering before you even see anything in the logs.
What’s the difference between BGP and OSPF?
OSPF is a link-state interior gateway protocol used inside one organization to find shortest paths based on link cost. BGP is a path-vector exterior gateway protocol used between organizations to exchange reachability and apply policy. OSPF cares about topology. BGP cares about which networks are reachable through which neighbors and what business rules govern that.
Can BGP run inside one company’s network?
Yes. iBGP runs between routers in the same AS. It’s common in large enterprises and service providers that need to carry BGP-learned routes across their internal network, especially when those routes came from external peers and need to reach internal customers.
What is AS path prepending?
AS path prepending is the practice of adding your own AS number to a route advertisement multiple times to make the path look longer than it really is. Because BGP prefers shorter AS paths, prepending makes the path less attractive to other networks. It’s a common tool for steering inbound traffic toward a preferred ingress link when you have more than one ISP.
How long does BGP take to converge?
Slower than OSPF. Convergence depends on the size of the routing table, the number of peers, and how many MRAI timers and route processing delays are in the path. At internet scale, convergence after a major change can take minutes. This is intentional. Fast reconvergence on the global table would cause more harm than good.
Do I need to learn BGP if I’m going for CCNA?
Not for the exam. For your career, probably yes, eventually. The CCNA gets you in the door. The first time someone hands you a ticket about an ISP failover that didn’t work, BGP is what you’ll be looking at.
Network Professional | CCNA Certified
Ashley Miller is a 35-year-old networking professional with a proven foundation in Cisco technologies. She is CCNA certified and currently advancing her expertise by working toward the Cisco Certified Network Professional (CCNP) certification. With a passion for designing and maintaining efficient, secure network infrastructures, Ashley brings both technical skill and real-world experience to every project.
