Understanding BGP and Why Route Leaks Cause Outages

What Is BGP?

The Border Gateway Protocol (BGP) is the routing protocol that holds the internet together. Every major network — ISPs, cloud providers, CDNs, enterprises — is an Autonomous System (AS), identified by a unique AS number (ASN). BGP is how these networks advertise reachability: "I can get you to 203.0.113.0/24, and here's the path to do it."

When you type a URL, your request crosses multiple ASes. BGP is what each router along the way uses to decide which AS to forward your packet through next.

Key BGP concepts

• Prefix: An IP address block, e.g. 192.0.2.0/24. Networks announce ownership of prefixes.
• AS path: The list of ASes a route advertisement has passed through. Routers use this to detect loops and choose paths.
• Route: A combination of a prefix and the attributes that describe how to reach it.
• Peer / transit: Two ASes can exchange routes as equals (peering) or one can pay the other to carry traffic (transit).

BGP is a protocol built on trust. When AS64500 announces it owns a prefix, its neighbors believe it — and propagate that announcement further.

What Is a Route Leak?

A route leak happens when an AS re-advertises routes it has learned from one neighbor to another neighbor in a way that violates policy or intent.

The classic scenario:

1. AS A (a content provider) announces 10.0.0.0/8 to AS B (a small ISP, its transit provider).
2. AS B is also peered with AS C (a large ISP).
3. AS B mistakenly re-announces 10.0.0.0/8 to AS C, as if B itself is the origin or a valid transit path.
4. AS C, seeing what looks like a shorter or more attractive path, starts routing traffic for 10.0.0.0/8 through AS B.
5. AS B was never meant to carry that traffic. It either drops it, lacks capacity, or forwards it incorrectly.

The result: traffic that was flowing fine suddenly takes a broken path. From the outside, the service looks down.

Route leaks are distinct from BGP hijacks, where a malicious or misconfigured AS announces a prefix it doesn't own at all. Leaks are usually accidental; hijacks may not be.

Why Leaks Cause Real Outages

BGP's preference for shorter AS paths means a leaked route often wins over the legitimate path, even globally. A single misconfigured router at a mid-tier ISP can redirect massive volumes of internet traffic.

This has happened repeatedly with major providers. In each case the underlying mechanism is the same: BGP believed the advertisement because the protocol has no built-in cryptographic verification of route ownership by default.

The impact compounds because:

• Traffic gets redirected to a network not provisioned to handle it, causing congestion or drops.
• Legitimate prefixes get de-preferred across many ASes simultaneously.
• Recovery requires the leaking AS to withdraw the routes, which propagates through BGP convergence — a process that can take several minutes.

Mitigations Worth Knowing

RPKI (Resource Public Key Infrastructure)

RPKI lets prefix owners sign Route Origin Authorizations (ROAs) — cryptographic records saying "AS64500 is authorized to announce 192.0.2.0/24". Routers that perform Route Origin Validation (ROV) can then reject or de-prefer RPKI-invalid announcements.

RPKI is the most widely deployable mitigation available today. If you operate an ASN, creating ROAs for your prefixes is low-effort and directly reduces your exposure to hijacks and some leak scenarios.

Filtering and IRR

Internet Routing Registries (IRRs) like RIPE's database or RADb allow networks to publish what prefixes they should be announcing. Responsible ISPs build prefix filters from IRR data, accepting only what a customer or peer claims to own.

BGPsec

BGPsec extends cryptographic signing to the full AS path, not just the origin. Adoption is minimal due to operational complexity, but it's the long-term direction the IETF has standardized.

AS path length and max-prefix limits

Simple but effective: configure your routers to reject routes with implausibly long AS paths, and set maximum prefix limits on BGP sessions so a runaway peer can't flood your table.

How Monitoring Helps You Detect the Impact

You usually won't see BGP events directly unless you have your own ASN and peering. What you will see is the downstream effect: your service becomes unreachable from parts of the internet.

This is where external monitoring from geographically distributed vantage points earns its keep. If checks from multiple regions start failing simultaneously while your infrastructure shows green, the problem is almost certainly in the network path — and BGP is a prime suspect. That asymmetric signal ("reachable from US-East, unreachable from EU and Asia-Pacific") is exactly the fingerprint a route leak or partial hijack leaves.

Key Takeaways

• BGP is a trust-based protocol — it accepts route advertisements without cryptographic proof by default.
• A route leak occurs when an AS re-advertises routes beyond their intended scope, redirecting or blackholing traffic.
• Leaks cause outages because BGP path selection may prefer the leaked route globally, and recovery depends on convergence time.
• RPKI + ROV is the most actionable mitigation for most organizations; create ROAs for your prefixes if you have an ASN.
• IRR filtering and max-prefix limits are complementary controls that responsible operators should configure.
• External uptime monitoring from multiple regions helps you distinguish a BGP-level incident from an application or infrastructure failure — the geographic pattern of failures is a key diagnostic signal.