PCCW Asia Segment Congestion Impacts US–Singapore Traffic

On February 26, 2025, traffic from the United States to Singapore experienced severe packet loss due to congestion within the PCCW Global backbone (Asia segment).

The incident occurred between 19:56 UTC and 21:43 UTC, during which packet loss peaked at approximately 60%.

PCCW Asia Backbone – Packet Loss Observation

Monitoring systems detected abnormal latency spikes followed by sustained high packet loss on transit paths routed through PCCW’s Asian backbone infrastructure.

Impact Scope

• US → Singapore traffic via PCCW transit severely degraded
• Packet loss peaked at ~60%
• Increased RTT and jitter observed
• Cloud services and real-time applications impacted
• Partial rerouting attempts seen during incident window

Services potentially affected included cloud workloads, financial trading links, CDN traffic, gaming services, and VoIP connections relying on PCCW transit for Asia-bound routing.

Technical Analysis

1. Symptoms Observed

• Sudden increase in packet discard rate at Asia backbone hops
• Latency variation beyond normal baselines
• No full BGP withdrawal — routes remained advertised
• Traffic continued flowing but with heavy degradation

2. Likely Cause

The pattern indicates backbone congestion rather than physical cable cut. Routes were not withdrawn, suggesting interfaces remained operational, but were likely saturated due to traffic surge or capacity imbalance.

Common triggers for this type of behavior include:

• Unexpected traffic spike
• Capacity planning misalignment
• Maintenance or partial backbone reconfiguration
• Failure on parallel path forcing traffic concentration

Timeline (UTC)

• 19:56 — Initial packet loss detected
• 20:05 — Loss exceeds 30%
• 20:22 — Peak loss ~60%
• 21:10 — Gradual recovery begins
• 21:43 — Metrics return to normal baseline

Conclusion

The February 26 congestion event highlights the importance of diversified transit strategy for intercontinental traffic. While no submarine cable failure was detected, backbone congestion within PCCW’s Asia segment caused substantial service degradation.

Operators are advised to:

• Implement multi-homed upstream design
• Monitor packet loss thresholds proactively
• Automate failover routing when loss exceeds SLA limits