Summary

  • On 2 October, CableStatus reported an average 177 ms on the Philippines–Singapore route, against a recent 53 ms baseline across 28 RIPE Atlas anchors. It cautioned that the measurements did not prove a physical fault or identify a cable.
  • InfiniVAN said its C2C Philippines–Singapore segment was restored at 12:48 p.m. on 3 October. Converge and Globe later described their own traffic and service states. The public record does not provide one shared, independently measured all-clear for every route or customer.

The useful question after the Philippines’ 2 October connectivity disruption is not simply when “the cable” came back. It is which thing a restoration notice says has returned: a measured path, a named cable segment, an operator’s traffic, or the service a customer buys.

The first public signal was an observation, not a repair report. PhilSTAR Tech said the University of the Philippines Diliman Network Helpdesk had been warned that an undersea segment between the Philippines and Singapore was offline. The helpdesk expected higher latency and intermittent performance, especially for overseas-hosted sites, and said effects could vary with each provider’s main links. CableStatus measured roughly 177 milliseconds on the route, compared with a recent baseline of about 53 milliseconds, using 28 RIPE Atlas anchors.

Its warning was explicit: those probes could not confirm a physical cable fault or name the affected system. GeoCables did not show a country-level outage at its latest check, though it showed varying dips among individual providers. A route sample, user reports and a national status label measure different things.

The next update came from an operator that could speak to a particular segment. On 3 October, InfiniVAN told PhilSTAR Tech that the C2C Philippines–Singapore segment had been restored and monitored since 12:48 p.m. The report also said the C2C Hong Kong link was under maintenance, limiting redundancy. That is a more specific service-status statement than a probe, but it still describes the operator’s view of those links; it is not a published inspection record for every international path used in the country.

Carrier statements then supplied a third clock. Converge said it stabilized service quickly by rerouting traffic, mostly through the Bifrost system. Globe said its mobile and broadband services were fully restored on Sunday, 4 October, after traffic resumed across the affected segment following reactivation of the Philippine landing station. Globe also said it was still working with international partners to determine the exact nature and location of the fault and would keep monitoring performance. These updates concern different networks and customer bases.

Their dates do not prove that one operator lagged another or that users experienced the same degradation.

That distinction matters to a company running workloads across Singapore-hosted applications, cloud regions, payment systems or collaboration services. A route can be reachable while latency remains elevated. A carrier can move traffic to an alternate path while the original segment is still unavailable. A network-level “normal” notice may not show whether a particular application, city or enterprise circuit met its own performance objective.

The public evidence has useful but narrow edges. It documents a measured latency spike at a defined set of probes, InfiniVAN’s report on one C2C segment, and carrier descriptions of rerouting or service normalization. It does not provide a common incident log, path-by-path capacity utilization, a final Reason for Outage report, or independent before-and-after measurements for each carrier’s customers. It does not establish the fault’s physical cause or exact location. No inference about sabotage or uniform nationwide impact follows from these reports.

The operational conclusion is simple: “restored” should name its object. A link-state update, a carrier network update and a customer service closeout are separate receipts. For this incident, the public reporting did not collapse them into one independently verified result.

Sources