TWAMP, Y.1564 and Y.1731: proving SLA on an open network

01TWAMP: active IP performance measurement

TWAMP, the Two-Way Active Measurement Protocol (IETF RFC 5357), measures the real performance of a live IP path by sending timestamped test packets between two endpoints and reading the results back. A session-sender originates the test traffic and a session-reflector bounces it, so the sender can compute round-trip delay, delay variation, and packet loss on the exact path a service rides. Because it is an open standard, an OcNOS endpoint interoperates with any conformant TWAMP sender or reflector.

OcNOS implements the full protocol on both sides. A router can be the control-client and session-sender that runs the tests, the server and session-reflector that answers them, or both at once, and it also supports one-way measurement where the metric of interest is one direction only.

What makes this useful for carriers is where OcNOS can run the measurement. TWAMP sessions run over plain MPLS transport and, importantly, inside the customer VPN:

• MPLS L3VPN and EVPN MPLS L3VPN, so the SLA is measured in the same VRF the customer's traffic uses.
• L3VPN over Segment Routing MPLS and L3VPN sobre SRv6, for operators who have moved transport to SR.
• Inter-VPN access interfaces, to measure across service boundaries.

Measuring inside the VPN matters because a path that looks healthy on the underlay can still miss its SLA for one tenant. TWAMP in the service context measures what the customer actually experiences.

02Y.1564: service activation testing

ITU-T Y.1564, the Service Activation Test (SAT), is run once when a service is turned up, to prove it meets its SLA before the customer is told it is live. It replaces the older RFC 2544 methodology for multi-service Ethernet because it tests all of a service's SLA parameters together rather than one at a time.

A Y.1564 test has two stages. The configuration test ramps traffic through the committed information rate (CIR) and excess information rate (EIR) to confirm the policer and shaper behave as provisioned. The performance test then soaks the service at CIR for a sustained period and verifies the four SLA metrics at once:

• Information rate (throughput at CIR and EIR),
• Frame Loss Ratio (FLR),
• Frame Transfer Delay (FTD),
• Frame Delay Variation (FDV).

OcNOS runs the Y.1564 SAT natively, so a turn-up technician can validate an Ethernet service to its SLA from the OcNOS CLI instead of dispatching a separate test set to each end.

03Y.1731: continuous Ethernet OAM

Where Y.1564 proves a service at birth, ITU-T Y.1731 Ethernet OAM watches it for life. Built on IEEE 802.1ag Connectivity Fault Management, it gives every Ethernet Virtual Connection its own always-on fault detection and performance monitoring. OcNOS implements the full toolkit:

04How the three fit together

The three standards are not alternatives. They are the assurance lifecycle of a service, and OcNOS runs all of them on the same router:

Together they let an operator sell a hard SLA and back it with measurement at both the Ethernet and IP layers, on open hardware, without a rack of proprietary probes.