Shop Applications News About Contact Book a meeting Connectivity guide Search GMTS CONNECT ↗ VesselNow ↗ GMount Network Status Login

← Back to news

Apple, Microsoft, Cloudflare: a week of outages. Is your business internet ready?

Share LinkedIn
Apple, Microsoft, Cloudflare: a week of outages. Is your business internet ready?
Dense bundle of blue, yellow and green network cables plugged into a rack-mounted Ethernet switch
Photo: Raysonho @ Open Grid Scheduler, CC0, via Wikimedia Commons.

It’s been a rough week for the internet. A global Microsoft 365 outage took Exchange Online email down around the world. Apple’s App Store, Apple Music and iCloud sign-in fell over after an internal update. Cloudflare logged an incident on two of its developer services. In the US, flights were grounded and banks went down.

The BBC reported that every affected company said its outage wasn’t a cyberattack, and experts found no link between them. As one put it, the infrastructure the world runs on is “changing faster than ever, and change is risk”.

For a business in Hobsonville, Hamilton or Hokitika, the lesson isn’t to panic about the cloud. It’s simpler than that: one connection, one cloud and one path is a single point of failure.

What actually went down (in plain English)

  • Microsoft 365, 12 November (US time; Friday morning, 13 November, in NZ). UC Today explains Microsoft’s preliminary report: a latent defect in a July update to the gateway servers in front of Exchange Online, made worse by a Teams failover and about 14% of capacity already out of action. At the worst point, more than 80% of those servers stopped responding. Teams meetings largely carried on. Email didn’t.
  • Apple, 17 November (US time; 18 November in NZ). Euronews reports Apple Music, the App Store and AppleCare had full outages and iCloud sign-in had problems. Apple blamed “an internal update”.
  • Cloudflare. Its status page shows a resolved incident with increased errors on Durable Objects and D1, two services developers build apps on.

None of these was your fibre, your router or your ISP. Your internet was fine. The services on the other end weren’t.

ISP, cloud or your own gear? How to tell fast

When staff shout “the internet’s down”, run through this before ringing anyone:

  1. Can you reach a few different sites? If Google, your bank and a news site all load, your internet is fine. The problem is the service.
  2. Check the vendor’s own status page. Microsoft 365, Apple and Cloudflare all publish one. Go there directly, not through a link in an email.
  3. Is it everyone, or one floor, one app, one device? One person points to a device. One app points to the service. The whole site points to the WAN.
  4. Look at the router. A good business router will tell you which WAN link is up, which is down, and when it switched.

Getting this right in five minutes saves an hour on hold to your ISP about a problem that isn’t theirs.

Dual-WAN: the part you can control

You can’t fix Microsoft. You can make sure the outages that are yours, like a digger through the fibre, a dead ONT or a provider fault, don’t stop your business.

The standard design we use is:

  • Fibre as the primary.
  • A second, independent path: 4G/5G cellular, Starlink, or both. “Independent” matters. A second fibre down the same trench isn’t a backup.
  • A router that switches automatically, so staff barely notice.

For yards, marina offices and depots, that usually means fibre plus cellular, with Starlink added where cell coverage is weak or the site is remote.

Bonding vs simple failover

These get mixed up a lot.

  • Failover switches to the backup when the primary drops. It’s simple and it works, but active sessions such as a video call or a card payment can drop during the switch.
  • Bonding uses several links at once. Peplink describes SpeedFusion as combining connections into a single tunnel, with Hot Failover moving traffic to healthy links so active sessions stay up, and WAN Smoothing sending duplicate packets to cover for loss.

Bonding isn’t just a Peplink idea any more. In November, Teltonika launched WANplex, its own SD-WAN bonding service, which shows where the industry is heading. On GMTS sites, our standard is Peplink SpeedFusion, managed in InControl.

Which do you need? If a 30-second blip is fine, failover will do. If you run VoIP, Teams calls, EFTPOS or remote access that can’t drop, look at bonding.

Monitoring, and a “test the failover” routine

A backup you’ve never tested is a hope, not a plan.

Failover checklist for business sites
– Primary and backup links are genuinely independent (different technology or provider)
– Router alerts you, by email or app, when a link fails or switches
– Someone owns the alerts and knows what to do
– Failover is tested on purpose, ideally quarterly: unplug the primary and watch
– Critical apps (phones, EFTPOS, remote access) are checked during the test
– A one-page runbook on the wall: who to call, which light means what
– Status pages for your key cloud services are bookmarked

When we commission a site, we fail the primary on purpose in front of you and write the runbook while we’re still on site. Then InControl keeps watching after we leave.

The bottom line

The big cloud outages will keep happening. Your job isn’t to prevent them. It’s to know in five minutes whether a problem is yours or theirs, and to make sure the problems that are yours never take the business offline.

Talk to the team

Want to know how your site would cope? Talk to the enterprise team or book a review. See business internet and failover, Peplink SD-WAN and managed networks.

We’re a certified supplier and installer, and we can support you or your preferred installer on site. Starlink airtime is also available with flexible plans and pool options.

Sources

← Back to news

Talk to GMTS ← Back to news