back to article Azure outages ripple across multiple dependent Microsoft services

Microsoft has reported two Azure service wobbles in as many days, including a disruption affecting Virtual Machine management ops yesterday and a Managed Identity for Azure resources outage in East US and West US regions today. According to Microsoft, today's Managed Identity for Azure resources issue impacted the East US and …

  1. IGotOut Silver badge

    Is MS365...

    ...the amount of outages it has?

  2. Rich 2 Silver badge

    Ha ha….

    Ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha……

    1. cookieMonster
      Mushroom

      Re: Ha ha….

      I’m joining in……

      ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha

      Use the cloud they said

      It’s cheaper they said

      It’s resilient they said

      ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha

      1. Michael Strorm Silver badge

        Sorry

        Knock knock.

        Who's there?

        Azure.

        Azure who?

        Azure cloud service gone down as well?

  3. An_Old_Dog Silver badge

    Circular Failure

    You cannot manage your cloud-based virtual machine when the virtual machine management service is running on a failing cloud-based virtual machine.

    Manage the cloud from outside of the cloud.

    1. Autonomous Mallard

      Re: Circular Failure

      There was an interesting article last year about Atlassian running a DR exercise and discovering their architecture had _tons_ of circular dependencies: https://www.theregister.com/2025/11/25/atlassian_dependency_migration/

      Atlassian remedied some of those issues by essentially turning the dependency graph into a DAG where possible. I've never seen a similar analysis for Azure, but I'd be surprised if they're much better.

      I don't think anyone knows how long a restart of these distributed systems would take after a total failure (e.g high-altitude nuclear detonations, Carrington event). Hopefully we'll eventually have plans similar to what utilities have for "black starts" of the power grid.

      1. druck Silver badge

        Re: Circular Failure

        I don't think anyone knows how long a restart of these distributed systems would take after a total failure (e.g high-altitude nuclear detonations, Carrington event).

        Or in the case of Azure; a door mouse farting.

  4. Autonomous Mallard
    Facepalm

    > The outage, which has been mitigated, "impacted dependent services such as... Azure Chaos Studio...

    The outage caused a failure in the outage simulator. Beautiful.

    1. Michael Strorm Silver badge
  5. Bebu sa Ware Silver badge
    Coat

    Appears Meteorology might be the appropriate discipline

    for the Azure and other clouds.

    Daily forcasts of the "weather" expected, broadcast on the news channels.

    Certainly the Amazonian butterflies have been busy recently. ;)

    1. breakfast Silver badge

      Re: Appears Meteorology might be the appropriate discipline

      I mean an Azure sky contains no clouds, so it makes sense.

      I just love that the general term for an Azure outage, "Blue Sky Of Death" works on so many levels.

  6. Anonymous Coward
    Anonymous Coward

    Figures

    We appear to have a fully compatible IT team, they've never heard of change control either.

    Not that Microsoft makes that easy - go ahead, try and preserve a Microsoft Purview DLP configuration in one go so you can roll back changes. Not possible.

  7. frankvw Silver badge
    Facepalm

    Perhaps...

    Maybe, just maybe, we are all (dare I say it) becoming just a tat too dependent on third party SPOFs hosted by overseas tech giants who care about nothing but their stranglehold on the market and their bottom lines? Cloudflare, AWS, Azure and what not have lately proven to be more of a liability than a boon to the IT using world. Or maybe that's just me.

    1. Zack Mollusc

      Re: Perhaps...

      It's just you. Those overseas tech giants are currently building agentic AI systems to oversee their operations and therefore the future will be entirely trouble-free.

  8. Pirate Peter

    Vibe coding strikes again??

    I have heard (off the record from someone who worked at MS ) the internal MS staff have to use AI tools for all coding, and not allowed to write code by hand

    so I wonder how much of this is relying on AI to generate the code for these updates and patches, then using AI tools to test it before deploying

    if it is the case, my answer would be that if they can't make it work reliably why should we use it

    I have already seen the mess one of our devs is making using various "AI coding tools" and how much longer it is taking to clean up when it all goes wrong, as it does more often than not

    1. Jellied Eel Silver badge

      Re: Vibe coding strikes again??

      I have already seen the mess one of our devs is making using various "AI coding tools" and how much longer it is taking to clean up when it all goes wrong, as it does more often than not.

      This seems like a rather big risk, ie developers turn into auditors, until that function is AI'd. AIs verify their own code (for a fee, naturally) and there's fewer people who can figure out what the code is or does on production systems. Which might then mean developers who've cut their teeth on debugging & fixing AI slop will be able to pick & choose their own contracts.

      1. Anonymous Coward
        Anonymous Coward

        It will be AI all the way down

        AI will be debugging and fixing AI generated production code. Humans will not get to see anything other than a sharepoint "black box" slide of inputs and outputs.

      2. retiredFool

        Re: Vibe coding strikes again??

        I would compare it to the chip industry except... Synthesis became a thing quite awhile ago. The logic blobs created were huge from some RTL code. No human was going to sort thru the blob to verify it. The difference was it was not an AI, but a highly reliable algo that did the transformation. And even with that, a logic sim is run on the output to validate, along with timing of course. Pity AI is not more deterministic.

  9. Taliesinawen

    Testing changes before deploying to production :o

    > Those same developers would likely have some advice for Microsoft on testing changes before deploying to production.

    How does one test a configuration change on a cloud platform. Seeing as it can't ever be turned off or rolled back to a previous instance.

    1. Anonymous Coward
      Anonymous Coward

      Re: Testing changes before deploying to production :o

      You set up a dev partition, and a test partition.

      What you do NOT normally do is roll it out into production and user your customers to test and debug, but that's what Microsoft's latest releases feel like.

      1. Excused Boots Silver badge
        Facepalm

        Re: Testing changes before deploying to production :o

        Every company has a test and dev facility to roll out the latest fad and see what happens

        A tiny minority also has a separate and isolated production facility.

  10. David Harper 1

    Your weekly reminder ...

    ... that "the cloud" is just somebody else's computers, over which you have no control.

  11. IGnatius T Foobar !

    I see the problem

    ...it's running on Microsoft software

    1. Anonymous Coward
      Anonymous Coward

      Re: I see the problem

      I think "running" is overstated.

      It's more limping along on broken crutches, taking the occasional break.

  12. Uh, Mike

    "Those same developers would likely have some advice for Microsoft on testing changes before deploying to production."

    Why revoke the enjoyment of being a crash test mannequin?

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon