CONCEPT Cited by 1 source
BGP path selection¶
BGP path selection is the deterministic tiebreaker sequence a BGP router uses to select the single best path for each prefix from among all candidate routes received from neighbors. Defined in RFC 4271 §9.1, the process evaluates attributes in a fixed priority order — each step only applies when all preceding steps result in a tie.
Selection ladder (simplified)¶
| Step | Attribute | Prefer |
|---|---|---|
| 1 | Local Preference | Highest |
| 2 | AS_PATH length | Shortest |
| 3 | ORIGIN | Lowest value (IGP=0 < EGP=1 < INCOMPLETE=2) |
| 4 | MED (Multi-Exit Discriminator) | Lowest |
| 5 | eBGP vs iBGP | eBGP preferred |
| 6 | IGP metric to next-hop | Lowest |
| 7 | Router ID | Lowest (final tiebreaker) |
Why ORIGIN matters disproportionately¶
Because Local Preference is typically set uniformly for routes of the same business-relationship class (customer / peer / transit), and AS_PATH length is often equal for competing routes from the same tier, ORIGIN frequently becomes the effective tiebreaker in practice. A single-value change (INCOMPLETE → IGP) can redirect traffic without any path-length manipulation — making it an attractive target for traffic engineering by transit providers seeking to attract more traffic (and revenue). (Source: sources/2026-07-24-cloudflare-bgp-origin-attribute-manipulation)
Seen in¶
- sources/2026-07-24-cloudflare-bgp-origin-attribute-manipulation — detailed analysis of how ORIGIN manipulation exploits step 3 of the ladder
- sources/2026-01-08-cloudflare-a-closer-look-at-a-bgp-anomaly-in-venezuela — AS_PATH prepending (step 2 manipulation) as an alternative traffic-engineering signal