Skip to main content
Blue Sky

When the Cloud Goes Dark: Why Owning Your Platform Still Matters

By Matt McCloskey, Founder and CEO, Blue Sky

Published · 5 min read

  • 2 days from the first failures to full service
  • 130,000+ user reports of problems
  • 12+ Microsoft 365 services degraded

Two Days Without Email

Colourful envelopes piled on an orange surface

At around 5pm on Thursday 12 November, Outlook stopped working. Not for one company or one region but for a large share of Microsoft 365 customers worldwide. Mailboxes would not open, mail queued up and anything that leaned on Exchange Online (Teams calendars, SharePoint, Copilot, Planner) went with it. Full service was not back until the morning of Saturday 14 November.

In the UK the impact landed hardest where it hurts most. Staff across hospitals, GP practices and community services lost access to NHS.net Connect. Some operations were cancelled, urgent referrals (including cancer referrals) were disrupted and some teams went back to paper records. Pharmacies warned that prescriptions could be delayed because they could not reach GPs. Schools, Ofqual and the Department for Education reported major disruption, and a government minister was left calling for "greater resilience" against technology failures.

Microsoft's explanation was a configuration change after an update that overwhelmed the process handling secure connections (GeekWire, Wikipedia). Not a cyberattack. Not an exotic failure. A routine change, made by someone else, in someone else's data centre, that took out email for millions of organisations at once.

The Real Problem Isn't the Outage

A wooden hourglass on a yellow background

Every platform fails eventually. Ours included. Anyone who tells you otherwise is selling something.

The problem is what happened next inside each affected organisation: nothing. There was nothing an NHS trust IT team, a council or a government department could do except refresh the Microsoft 365 status page and wait. No engineer to call who could look at their servers. No option to fail over. No ability to prioritise their own critical services over anyone else's.

Thousands of capable IT teams were reduced to spectators.

That is the shift that has happened quietly over the last decade. Organisations used to own their platforms and rent the infrastructure. Increasingly they rent everything, and with it they have handed over control of when things break, how they are fixed and in what order they come back.

Concentration Risk Is a Choice

Eggs in a single white basket

When most of the UK public sector runs email and identity on the same service, a single bad change becomes a national incident (Computer Weekly). That is not a failure of any one IT department. It is the predictable result of everyone making the same individually sensible decision.

We would never accept that level of concentration in power, water or payments. Yet for the system that carries referrals, procurement, case work and almost every internal decision, we largely have.

What Sovereignty Actually Means

"Sovereign" gets used loosely, so here is what we mean by it:

  • Where your data lives In UK data centres, under UK law, not subject to foreign legislation such as the US CLOUD Act.
  • Who operates it A UK team you can name, phone and hold to account, rather than a status page and a queue.
  • Who decides on change Updates and configuration changes scheduled around your organisation, not rolled out to millions of tenants at once.
  • Whether you can leave Your data in open, portable formats, with a realistic exit plan, rather than being locked into one vendor's roadmap and pricing.

None of this means rejecting the public cloud. It means being deliberate about which services you hand over entirely and which you keep under your own control.

A Practical Approach

For most organisations the answer is not "all in" or "all out". It is a considered mix:

  1. Identify the services that cannot stop. Clinical communications, case management, critical workflows. Ask honestly what happens if each one is unavailable for 36 hours.
  2. Avoid a single point of failure for those services. That might mean hosting them separately, on your own platform or with a specialist UK provider, so one vendor's bad day does not take everything down together.
  3. Keep the skills in-house to understand your own estate. Outsourcing operation is fine. Outsourcing understanding is not.
  4. Test your exit. If you could not move a service elsewhere within a reasonable time, you do not really own it.

Where Blue Sky Fits

Three coloured keys

For more than 20 years Blue Sky has run business-critical HCL Domino applications for NHS organisations, councils and central government, hosted in UK data centres and operated by a UK team. We are ISO 27001 and ISO 9001 certified and available through G-Cloud 15.

Our customers know who looks after their systems. When something needs attention they speak to the engineers who run their servers, and changes happen on a schedule agreed with them.

We supply Microsoft 365 too, and for plenty of workloads it is the right answer. The point is choosing where it fits, not defaulting to it for everything.

The Question to Ask This Week

While the outage is still fresh, it is worth asking a simple question at your next leadership or IT meeting:

If this happened again tomorrow, which of our critical services would we still control?

If the honest answer is "none of them", we would be happy to talk. Find out more about our Domino hosting or get in touch.

Talk to us about your critical services

About the author

Matt McCloskey, Founder and CEO, Blue Sky

Matt founded Blue Sky and has spent over 20 years hosting and managing HCL Domino for UK public sector and regulated organisations.

Matt McCloskey on LinkedIn

Accredited and certified