Repository navigation
Conversation
Toning down "purpose" of AS-SETs into "common use case", and dropping ROUTE/ROUTE6 objects as "the cornerstone" of IRR-based filtering. Extended reasoning: My operational practice managing IRR objects for the PEERING Research Testbed (AS47065) is that ROUTE-SETs are significantly more flexible and easier to manage than the AS-SETs and ROUTE/ROUTE6 combo widely employed by operators today. This RFC is an opportunity to steer operators towards ROUTE-SETs and possibly improve operational practice. ROUTE-SETs allow precise specification of prefixes via RPSL prefix-range operators, and an AS can define multiple ROUTE-SETs for different peering relationships. ROUTE/ROUTE6 objects are not hierarchical, have few contraints on who can create them, and do not support range operators. As AS added as a member of an AS-SET contributes the exact same of routes to a prefix, which implies that AS-SETs cannot define different sets of prefixes better suited to different peering relationships. RPSL allows operators to specify import and export rules using the following syntax (in aut-num objects): mp-import: from <PEER> import <FILTER> mp-export: to <PEER> export <FILTER> I would argue that AS-SETs are better fit for specifying the <PEER> part of import/export rules (which is usually just an ASN), and that ROUTE-SETs are a better fit for the <FILTER> part of import/export rule. The only limitation I can think of ROUTE-SETs relative to the AS-SET plus ROUTE combo is that ROUTE-SETs alone do not allow AS-path-based filters that check that the originator of an announcement is within a given AS-SET. This does not seem like a significant loss: (1) I would guess that few operators configure AS-path-based filters, and (2) the RFC should recommend that operators deploy ROV for origin validation instead of ROUTE/ROUTE6 objects.
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
The ROUTE-SET security limitations and hierarchical authorization behavior need correction.
Review effort: Balanced
Findings: 1
Open (2)
What changed in this PR
Adds ROUTE-SET guidance for generating and managing IRR-based BGP filters.
Changes:
- Documents ROUTE-SET prefix ranges and hierarchy.
- Reorganizes filter management guidance.
- Extends hierarchical object recommendations.
| File | Description |
|---|---|
draft-ietf-grow-routing-ops-sec-inform.xml |
Adds ROUTE-SET operational and security guidance. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
| In 2022, an attack was observed where a malicious party created an empty AS-SET for a large transit provider (see <xref target="NLNOG-22" format="default"/>). | ||
| As it was created in an RIR database often taking precedence over other IRR sources, several ASes imported this empty AS-SET, and hence filtered all prefixes advertised by this transit provider. | ||
| To mitigate this issue, hierarchical AS-SETs and ROUTE-SETs reside in the IRR of the RIR and explicitly list the ASN to which they pertain, e.g., AS65536:AS-EXAMPLE or AS65536:RS-EXAMPLE. | ||
| In RIR-operated IRRs, only the maintainer of the ASN (or the corresponding aut-num object) before the colon is authorized to create and modify these objects, preventing unauthorized creation or spoofing. |
There was a problem hiding this comment.
@ichdasich Thoughts on whether we want to get into this level of detail? If you think it's worth it, I'll update the PR with the below (I find the AI phrasing a bit convoluted):
In RIR-operated IRRs, only the maintainer of the ASN (or the corresponding aut-num object) before the colon is authorized to create these objects, preventing unauthorized creation or spoofing. Subsequent modification is possible only by the owning maintainer or maintainers authorized by the
mnt-byfield.


Reasoning:
My operational practice managing IRR objects for the PEERING Research Testbed (AS47065) is that ROUTE-SETs are significantly more flexible and easier to manage than the AS-SETs and ROUTE/ROUTE6 combo widely employed by operators today. This RFC is an opportunity to steer operators towards ROUTE-SETs and possibly improve operational practice.
ROUTE-SETs allow precise specification of prefixes via RPSL prefix-range operators, and an AS can define multiple ROUTE-SETs for different peering relationships. ROUTE/ROUTE6 objects are not hierarchical, have few contraints on who can create them, and do not support range operators. As AS added as a member of an AS-SET contributes the exact same of routes to a prefix filter, which implies that AS-SETs cannot define different sets of prefixes suited to different peering relationships.
RPSL allows operators to specify import and export rules using the following syntax (in aut-num objects):
I would argue that AS-SETs are better fit for specifying the
<PEER>part of import/export rules (which is usually just an ASN), and that ROUTE-SETs are a better fit for the<FILTER>part of import/export rule.The only limitation I can think of ROUTE-SETs relative to the AS-SET plus ROUTE combo is that ROUTE-SETs alone do not allow AS-path-based filters that check that the originator of an announcement is within a given AS-SET. This does not seem like a significant loss: (1) I would guess that few operators configure AS-path-based filters, and (2) the RFC could recommend that operators deploy ROV for origin validation instead of ROUTE/ROUTE6 objects.