BGP Route Leaks and Hijacks: What They Are and How to Monitor

BGP Route Leaks and Hijacks: What They Are and How to Monitor

The global internet holds together because independent networks exchange reachability information via the Border Gateway Protocol (BGP). When policies and filters are correct, traffic follows predictable, efficient paths. When they are not, small mistakes—or hostile intent—can break that predictability and send packets on detours that hurt performance, availability, and security.

Two failure modes dominate these incidents: route leaks and route hijacks. Leaks are usually accidental policy violations that turn a non-transit network into transit. Hijacks are unauthorized originations of someone else’s prefixes. Both distort the control plane and can quickly impact the data plane by shifting traffic toward the wrong place.

You don’t need a global view to protect your users, but you do need timely signals that something has gone off script. That’s where cryptographic origin validation (RPKI), strict policy filtering, and independent route monitors work together. The goal is simple: detect and block bad announcements before users notice timeouts, latency spikes, or broken sessions.

Understanding BGP and Its Trust Model

BGP lets Autonomous Systems (ASes) exchange routes for IP prefixes. Key attributes influence path selection, including LOCAL_PREF, AS_PATH length, MED, and prefix specificity. Because most routers prefer more-specific prefixes, even a single more-specific false route can siphon traffic away from the correct origin. There’s no central authority enforcing truth in BGP; trust and bilateral policy rule the day.

What Is a Route Leak?

A route leak is the propagation of routes beyond their intended scope, violating business relationships or policy. The classic case: a customer learns full routes from one provider and unintentionally advertises them to another provider, acting as unauthorized transit. Leaks often create suboptimal routing (longer paths or congested detours) and may trigger widespread preference for the leaked path if it appears shorter or more specific.

Typical Causes of Leaks

Root causes are mostly operational: missing outbound filters, incorrect import/export policies, overly broad route-maps, not tagging routes with communities for containment, or failing to set max-prefix limits. Multi-homed customers without careful policy can leak provider-learned routes to peers. Hardware and software bugs, while rarer, can also mis-handle communities or attributes and cause leaks.

What Is a Route Hijack?

A hijack occurs when an AS originates a prefix it does not legitimately control. Sometimes it is a configuration mistake (fat-fingered prefix lists); sometimes it is intentional to intercept or blackhole traffic. Because origin is a critical trust signal, unauthorized originations can pull significant traffic if they look more attractive—shorter AS_PATH or more-specific prefix.

Consequences of Hijacks

Hijacks can cause outright outages (blackholing) or subtle interception (man-in-the-middle). They can degrade performance, break TLS stapling and certificate validation paths if traffic is rerouted through middleboxes, or disrupt CDN mapping and geolocation. Even brief hijacks can cause session resets and packet loss that translate into user-visible errors and financial impact.

How These Incidents Break Things

On the control plane, leaks and hijacks add incorrect entries to the routing table. On the data plane, packets follow those entries, landing on routers that lack state, capacity, or the real service. The symptoms: asymmetric routing, sudden latency jumps, increased retransmits, and timeouts. Because BGP convergence is slow relative to application expectations, even a few minutes of bad propagation can produce a large support spike.

Monitoring and Early Detection

External visibility is crucial because you rarely see the whole internet from a single AS. Effective detection combines three views: what you think you announce (your router configs and IRR/RPKI data), what the rest of the world sees (route collectors, looking glasses, streaming BGP feeds), and what users experience (synthetic probes). Your monitoring should alert on unauthorized origins, unexpected more-specifics, suspicious AS_PATHs, and route volume anomalies.

Signals Worth Alerting on

Practical triggers include: a prefix you own is seen with an unfamiliar origin AS; a more-specific of your space appears anywhere; sudden AS_PATH changes involving non-neighbor ASNs; excessive announcements from a customer; and invalid RPKI state (INVALID/UNKNOWN where you expect VALID). For latency-sensitive apps, augment control-plane alerts with synthetic transactions to catch impact fast.

RPKI and Route Origin Validation

RPKI lets holders of number resources publish Route Origin Authorizations (ROAs) that bind a prefix to an authorized origin AS and a maximum prefix length. Routers that perform Route Origin Validation (ROV) can mark announcements as VALID, INVALID, or UNKNOWN. Dropping INVALID routes prevents many origin hijacks and some kinds of leaks. To get value, publish ROAs for all originated prefixes, keep max-length tight to block accidental more-specifics, and ensure your edge routers or route servers validate.

Limits and Pitfalls of RPKI

RPKI does not verify the entire AS_PATH; a VALID origin can still be part of a leak. Misconfigured ROAs (too-narrow max-length or wrong origin) can cause self-inflicted reachability loss. Operationally, you must maintain ROAs in sync with provisioning, de-aggregation, and anycast changes. Still, with careful process, the risk of operational mishaps is smaller than the risk of unvalidated originations.

Policy Hygiene and Filtering

Good hygiene blocks bad events from spreading. Enforce prefix-lists on customer sessions; accept only the customer’s documented space and expected ASNs. Apply max-prefix limits with sane thresholds. Use well-scoped BGP communities to control export to providers and peers. Where possible, use peer-locking patterns to prevent exporting provider-learned routes to other providers. Keep Internet Routing Registry (IRR) objects current if you use IRR filtering alongside RPKI.

Operational Playbooks That Work

When an alert fires, confirm the anomaly with at least two vantage points, then decide whether to withdraw your more-specifics (to fight de-aggregation), announce covering prefixes from additional locations, or contact offending parties via NOC channels and peering coordinators. Prepare templated communications for upstream tickets and customer status pages. After stabilization, perform a post-incident review to tighten ROAs, filters, and alert thresholds.

Testing Your Defenses

Run staged drills: intentionally withdraw and re-announce test prefixes, validate ROV behavior, and confirm that monitors detect unexpected origins and more-specifics. Track time-to-detect and time-to-mitigate. Integrate alerting with on-call tooling so the right people see anomalies in minutes, not hours.

IPv6 Considerations

All of the above applies to IPv6, with the added operational burden of larger address space and more frequent use of de-aggregation. Ensure ROAs exist for IPv6 prefixes, max-lengths are carefully set, and filters cover v6 sessions with the same rigor as v4. Don’t assume your vendors treat v6 attributes identically—verify with lab tests.

What Actually Reduces Risk

No single control is sufficient, but together they change the odds: publish and maintain ROAs; enable validation on edges; filter customers strictly; monitor your prefixes from multiple external collectors; set max-prefix and damping where appropriate; and keep an operational runbook for leaks and hijacks. That layered approach shortens incidents and limits blast radius.

BGP Route Leaks and Hijacks (FAQ)

Use external route monitors and run an ASN Lookup on suspicious announcements to confirm the seen origin differs from your authorized AS.

Validate that export policies are intact and verify the customer is only advertising allowed space; remote vantage points and a fast DNS Lookup Tool can also reveal misrouted resolvers during leaks.

Combine synthetic probes from diverse regions with session logs and geodata; when in doubt, visualize affected areas on an IP Geolocation Map to spot unexpected detours.

Yes, if your ROAs set an appropriate max-length, validators can mark unauthorized more-specifics as invalid, similar in spirit to checking “who is allowed to originate what” like a targeted Reverse DNS Lookup checks name-to-IP integrity.

From a test host, confirm connectivity and verify the egress address with a quick What Is My IP Address check, then compare against your expected ranges.

Yes, publish ROAs and enable validation on v6 sessions; you can also sanity-check global reachability with an IPv6 Test from multiple vantage points.

They cap the number of accepted routes on a session, so if a peer leaks full tables inadvertently, the session trips the limit and protects your routers from overload.

Start by publishing accurate ROAs for your own prefixes and tighten customer prefix-lists; even partial deployment reduces blast radius and speeds incident triage.