Topic
Network-resource Evidence
Within the Topic facet, Network-resource Evidence topic intelligence connects articles that share a specific subject, signal focus, or monitoring theme. The page gives readers a richer path through related reporting, source evidence, market actors, and infrastructure implications, with enough context to understand why the topic matters across company movements, governance decisions, regional exposure, and operational risk. Readers can compare recurring signals, affected organisations, public evidence, market context, service continuity, procurement, competition, compliance, and strategic planning questions behind the subject instead of stopping at a thin list of matching articles. It explains what the topic covers, which infrastructure actors or policies are involved, what evidence supports the coverage, and why the subject may matter for operators, customers, investors, and policy readers.
History
Four Bytes Put a Message in TCP. They Did Not Rebuild the OSI Network: RFC 1006’s TPKT Boundary
TCP can deliver every byte in order and still decline to say where one application-shaped entity ends. In 1987, RFC 1006 addressed that awkward fact for ISO Transport over TCP with a remarkably small intervention: put four octets before a TPDU, then make the receiver count. The…
CASE FILE
The Legacy Code Point Reached the Client. It Did Not Reauthorise the Server: RFC 9963
An IANA code point can look like a broad permission slip when it appears in a migration review. RFC 9963 is deliberately narrower. It gives TLS 1.3 three legacy RSASSA-PKCS1-v1_5 signature values, but only for a client proving possession of a client-certificate key after a server…
CASE FILE
The Reverse Socket Arrived. The Device Had Not Yet Identified Itself: RFC 10011 and RESTCONF Call Home
A controller can receive a connection from the direction it expected, on a listener it deliberately opened, after a device has been configured to call home. That is useful evidence. It is not yet the same thing as knowing which device has arrived, establishing a RESTCONF session…
History
The Token Found the Connection. It Did Not Admit the Subflow: MPTCP's MP_JOIN Boundary
One new TCP SYN can propose to become part of a connection that began somewhere else, on another address pair, minutes earlier. That is the useful oddity in Multipath TCP. It is also where a casual description—“the session moved”—loses the decisions that keep continuity from…
CASE FILE
The Quantum-Safe Key Was Added. The Classical Recipient Still Opened the Mail: RFC 9980
RFC 9980 gives OpenPGP a post-quantum vocabulary, including composite encryption keys that combine ML-KEM with X25519 or X448. That is an important interoperability step. It is also easy to overstate. A message can carry a genuinely PQ/T-capable recipient key and still be…
IETF
Weiqiang Cheng and the SRv6 Locator Lease That Still Needed a Route
The DHCPv6 log can show a valid Reply, a locator, an IAID and two working clocks while the route table still has nothing to say. RFC 10038 makes that gap explicit: assignment is one controlled event; reachability must be created, distributed and observed through separate systems.
Story
LACNIC and the economics of transition architecture beyond RIRs
A transition test built around functions, records and authority—not around the fantasy that scarce network resources can govern themselves.
History
The Banner Marked the Screen. It Did Not Make the Policy: RFC 933 and Telnet's Display Boundary
A security banner could be sent once and kept visible by the workstation instead of being mixed into every screenful of application output. RFC 933 made that presentation bargain explicit—and left the security level, access decision and truth of the label outside the banner…
CASE FILE
The Model Could Explain the Network. It Could Not Authorize the Change: NEMOPS and the Boundary Between Visibility and Control
The NEMOPS workshop report asks for better models, observability, tooling and verification in network management. Those are valuable improvements in what an operator can see and test. They do not decide whose service may be changed, which risk is acceptable, or who bears the cost…
CASE FILE
The Fast Packet Kept the Session Up. It Did Not Authorise the Change: RFC 9985
RFC 9985 offers a disciplined answer to an awkward operational fact: a protocol can need frequent authenticated liveness packets at a rate that makes identical expensive authentication hard to sustain. Its answer is a split, not a blank cheque. Stronger-in-cost authentication…
History
The NAK Rejected the Port, Not the Packet: RFC 938 and the Boundary Between Delivery and Dispatch
In an experimental 1985 Internet protocol, a receiver could say two things at once: the next packet number had been received in sequence, and the local port named in that packet was not known. RFC 938 called that reply `PORT NAK`. Its precision is a useful warning against…
CASE FILE
An OAM Requirement Was Written Down. It Did Not Prove a Working Diagnostic: RFC 9974 and BIER Evidence
A requirements catalogue can sharpen an engineering question. It cannot, on its own, answer whether a particular network can run the test, what the test observed, or who may act on it.
Story
RIPE’s RPKI Key Plan Is Not Yet a ROA Scope Receipt
RIPE NCC’s plan to move RPKI API keys toward OpenID Connect is a statement about a future access mechanism. It is not yet a public way to reconstruct which resource authority and which state-changing action stood behind an individual ROA change.
CASE FILE
The Signed Time Saved a Transfer. It Did Not Prove a Moment: RPKI, RFC 9589 and the Switchover Boundary
A RPKI relying party can keep validating useful repository material after its preferred delivery path fails without pretending that a signed timestamp is a trustworthy clock. RFC 9589 makes that distinction operational: one CMS attribute can align filesystem comparison across…
History
The UUID Crossed the Connection. Authority Stayed Put: RFC 927 and the Double-Login Bargain
A TAC could check a name and password once, then send a four-octet user number to the next host. RFC 927's useful restraint lay in what did not travel with that number: the target still decided whether the first host's authentication was good enough for its own service.
Story
AFRINIC’s DBWG Shows 46 Open Items. A Count Is Not a State Path
Forty-six is a number that asks to be over-read. Put it next to the word Open and it can be made to sound like a verdict on an institution, a team or a service. AFRINIC’s public Database Working Group overview does not warrant that verdict. On the frozen 31 August 2026 capture…
CASE FILE
Four Thousand Was a Prefix Claim, Not a Rate-Limit Mandate: RFC 9977 and the Evidence Boundary
A prefix file can make an address range easier to group. It cannot convert a count of end sites into a command to relax a local risk control.
Story
LACNIC’s Bogon Guide Has Three Different Data Clocks.
A BGP filter can reject a route without saying why that route was rejected. Treating every rejected route as one security statistic may be convenient for capacity reporting, but it is too coarse to explain what the underlying evidence meant at the moment of the decision.
History
The Branch Said “Nobody Below”: RFC 1075 and DVMRP’s Expiring Non-Membership Report
In 1988, experimental multicast routing had to decide when not to send a group’s traffic down a branch. RFC 1075 offered a deliberately limited answer: a downstream router could say that it had no descendant members for one multicast group, and its upstream neighbor could stop…
History
The Keystroke Cost Four Octets: How RFC 916 Let State Replace Repetition
One typed character once faced a seven-octet packet around it. RFC 916 cut that packet to four octets by moving the character into the header position that would otherwise say how long the data was. The saving worked only because the connection had already made the missing facts…