Telco RAN DU reference design specification KPI test results for OpenShift 4.21

Updated

The Red Hat Telco QE team has validated the following KPI results on single-node OpenShift clusters that are configured with the telco RAN DU 4.21 reference design specification.

All results are taken from 12-hour test runs.

Realtime kernel KPI test
Workload nodes running the realtime kernel are validated to these performance KPIs:

  • No oslat samples greater than 20 µs.
  • 99.9999% of cyclictest samples are less than 10 µs. No samples above 20 µs.

Non-realtime kernel KPI test
The non-realtime kernel can be run with a reduced predictable latency target.
The following performance KPIs are validated:

  • No oslat samples above 20 µs.
    NOTE: cyclictest tests are not applicable for non-realtime systems.

RFC2544 KPI test

  • Zero packet loss at 99.9% line rate (25 Gbps) for 512 byte frames.
  • Packet latency is less than 30 µs for 100% of packets, measured at 80% line rate for 512 byte frames over a 12 hour test period.
    NOTE: The test application is the DPDK testpmd utility.

PTP network synchronization

  • The time offset for boundary and ordinary clock configurations, as measured at the follower port on the Intel E810-XXVDA4 (Salem Channel) NIC, is less than 100 ns, as indicated by the openshift_ptp_offset_ns metric.
Category
Components
Article Type