PTP — Precision Time Protocol
Sub-microsecond accuracy via hardware timestamping. For core banking switches, payment gateways, and hypervisor clusters.
PTP (IEEE 1588v2) synchronises the hardware clock of the network interface directly — eliminating the non-linear drift and stutter that occurs in virtualised environments under CPU load. Where NTS corrects clock drift at the software level, PTP anchors the hardware clock.
Why hypervisors need PTP
In a distributed data centre architecture, the risk to high availability is not inaccuracy — it is inconsistency. When hypervisors across a cluster drift relative to each other during peak load, distributed database clusters lose their single sequence of truth. During DR failover, a split-brain scenario can corrupt transaction ordering. PTP anchors every hypervisor across the estate to the same hardware-disciplined domestic source — the virtualisation fabric moves as one.
How it works
- Sub-microsecond accuracy — hardware timestamping at the NIC level, bypassing OS scheduling jitter entirely
- IEEE 1588v2 Grandmaster delivery — CNX provisions a unicast grandmaster feed; a boundary clock at each data centre site distributes the signal to hypervisors and core switches over the local network
- Designed for core payment switches and hypervisor estates — not a general-purpose service
Deployment
CNX provisions the unicast grandmaster feed. The institution's network team installs a boundary clock at each data centre site, which distributes the grandmaster signal to hypervisors and core switches over the local network.
Service levels
| Precision | <1 µs to UTC (real-time hardware) |
| Availability | 99.99% monthly |
| Incident response | 30-minute acknowledgement, 24/7/365 NOC |
| Atomic holdover | <1 µs/day drift during total GNSS or fibre outage |