Perth Subsea Cable Repairs

No internet-wide outage today. The major cloud, CDN and platform providers are reporting normal operations, and the notable activity is regional or long-running rather than global.

**Subsea cables off Perth.** Repair work on the Indigo West and Indigo Central cables continues, with the marine operator's estimated completion date falling today. Both systems took shunt faults in close succession over the weekend of 8–9 August, inside Australia's federally declared Submarine Cable Protection Zone, and the [Australian Federal Police are assessing a report](itnews.com.au) in connection with the faults. Traffic on the Perth–Singapore route has been rerouted around the breaks, so the practical effect has been reduced capacity and added latency rather than loss of connectivity. Timelines remain subject to weather and vessel logistics. ([SubTel Forum](subtelforum.com))

**AWS Middle East regions.** me-south-1 (Bahrain) remains unavailable and me-central-1 (UAE) is still unable to reliably support customer applications, following physical damage to those facilities earlier this year. Recovery has been measured in months, and AWS continues to advise affected customers to run workloads in other regions. This is a long-standing condition rather than a new event — it is also why some third-party trackers show AWS in a persistent warning state while every other AWS region operates normally.

**Twilio.** Regional carrier problems are still open: voice call failures between some Twilio numbers and Brazilian networks (recovering since 25 August), and SMS delivery delays to I.C.E subscribers in Costa Rica and Telefónica subscribers in El Salvador. Bulk Export is degraded. ([status.twilio.com](status.twilio.com))

**Slack.** An issue retrieving historical messages in workspaces with custom data retention policies has been open since 13 August. Login, connectivity, files, huddles and search are all operational. ([slack-status.com](slack-status.com))

**Core infrastructure is green.** AWS globally, Google Cloud, Microsoft Azure, Cloudflare, Akamai, GitHub, OpenAI, Anthropic and Discord all report normal service. Cloudflare has scheduled datacenter maintenance in progress in Chicago, Montréal and Hamburg, with a wider maintenance window running 27 August to 2 September. Akamai is monitoring a reporting-data issue in its Traffic by Hostname report after deploying a fix on 25 August. There is no sign of a broad routing, DNS or root-server problem.

**Resolved in the past 72 hours.** GitHub Actions start delays on 24 August, affecting about 3.8% of runs for 31 minutes; elevated model errors and intermittent login failures at Anthropic on 24–25 August; a DigitalOcean cloud control panel and API disruption on 24–25 August; and several short Cloudflare incidents in Los Angeles and Chicago on 24–25 August.

**Elsewhere.** Small, localized report spikes appeared for Sky Broadband in the UK, NTT Docomo in Japan and Indosat IM3 in Indonesia in the early hours of 26 August, and for AT&T on 25 August. Facebook drew a modest volume of user reports, mostly about ad delivery and slow loading, with no acknowledged Meta incident. The Internet Society counts 11 government-ordered internet shutdowns in force worldwide, including the ongoing blackout in Iran.

**Coverage note.** The Heroku, Stripe and Azure DevOps status pages did not render for automated retrieval this run, so those three are unverified; no independent signal suggested problems at any of them.