back to article Want to provision a new VM on Azure? Get in line

Multiple Microsoft customers have for the past nineteen hours been unable to provision new virtual machines in Azure's UK West and UK South regions. Although there has been no formal notification on the Azure status page, The Register has seen a copy of a Microsoft issue summary report indicating issues since Wednesday morning …

  1. Korev Silver badge
    Joke

    Have they thought about bursting into AWS?

  2. Anonymous Coward
    Anonymous Coward

    "The firm normally de-provisions Azure VMs overnight to save cost and restart them in the morning at 6.30am"

    Sorry but LOL

    1. Alistair 1

      Why LOL? Seems sensible to me?

      1. Anonymous Coward
        Anonymous Coward

        Because of this very reason.

        Maybe sensible for non-critical systems, but for the phone system, that's what made me LOL

        Do you turn off critical systems at night to save power? It does save money, but not a sensible thing to do.

        1. Alistair 1

          Depends what you define as critical & how much you reserve capacity etc. But sure, I turn servers off overnight, otherwise my bill would be 3x higher. Seems pretty sensible to me (but then I'm an AWS customer, rather than Azure...)

          1. Anonymous Coward
            Anonymous Coward

            Re:sensible

            I'll bite...

            I can buy the sensible cost savings as long as you can quantify the cost of having an unavailable phone system. And moaning about not having it on twitter doesn't make you money. I guess it saves the awkward conversations with customers that you convinced to use cloud to...

            As people using cloud services during outages or system faults have discovered before - when something goes wrong, what is the backup plan and does it actually work?

    2. Anonymous Coward
      Anonymous Coward

      Who needs phones anyway? A total phone system outage for a day doesn't kill a healthy business.

      While basically cancelling your phone line every night and buying it again in the morning does seem like a balmy move, the real LOL lies in their inflexibility to use the cloud responsibly.

      Why don't they just deploy their backup system in another Azure DC?

  3. JMiles

    Cloud is meant to be the new jewel in the Microsoft crown. What gives? Google & AWS don't have this sort of issue.

    Part of me thinks Azure is very much a 'one foot in the door' endeavour by Microsoft. If they can figure out how to white-label AWS it would not surprise me one bit.

  4. micahjj

    More than just VM's

    This issue is also affecting Batch Services as batch services spin up compute under the bonnet. We have P1's raised under enterprise support and have only 1 message sent at 18:00 yesterday.

    Really poor stuff from Microsoft. If you try to create a new vm via the portal at the moment you get an error saying that services are unavailable in the region, and prompted to use another data centre.

    A colleague with another consultancy I know also have another 2 P1's raised with little or no feedback either.

    Looks like amateur hour in the UKWest datacentre this week....

  5. Ken Moorhouse Silver badge

    Memory Leaks taken to a whole new level

    Switching the datacentre off and switching it back on again resolves the majority of issues.

  6. Anonymous Coward
    Anonymous Coward

    It is raining, it is pouring

    and microsoft is snoring.

    Sorry, could not resist.

    I hope that some of the PHB's piling into 'Cloud' as the solution to everything will learn from this but who am I kidding, they won't.

    We are Doomed I tell ye, doomed.

    Putting all your eggs in one basket is always the best solution....

    1. Captain DaFt

      Re: It is raining, it is pouring

      Putting all your eggs in one basket is always the best solution....

      Oh, we've learned from the past.

      Now we put all the eggs from several farms in one basket. and leave it out in the middle of the (information) highway. Safe as houses, of course!

  7. Dominion

    I love the irony that yet again these stories of cloud failures are sponsored by.....

    Sponsored: Ensuring end to end performance of a cloud phone system

    Sponsored: Buyers guide to cloud phone systems

    Sponsored: Getting the most value from cloud phone system

  8. Kicker of Metaphorical Cats
    Coat

    Must be BS

    "...the customer, speaking on condition of anonymity, told The Register."

    Come on now, if the US news media has taught us anything it is that anonymous sources = fake news. I thought this rag was better than that.

    Sorry, could not resist.

  9. Anonymous Coward
    Anonymous Coward

    The Cloud...

    Other peoples computers they don't have.

    1. Inventor of the Marmite Laser Silver badge

      The Cloud...

      Wherein lies Cuckoo Land

      1. TheVogon Silver badge

        Re: The Cloud...

        Can sometimes evaporate...

  10. Slabfondler
    FAIL

    One would think...

    a massive hosting provider would have monitoring in place to keep track of capacity, and ramp things up when usage gets to a critical level.

    1. Anonymous Coward
      Anonymous Coward

      Re: One would think...

      "a massive hosting provider would have monitoring in place to keep track of capacity, and ramp things up when usage gets to a critical level."

      Of course within reason. But if a lot of people decide to migrate a few thousand VMs on the same day, or someone decides to pay to spin up tens of thousands of new ones it could conceivably out pace reasonable planning. Or there could have been a significant failure - e.g. a complete container, etc. Microsoft's containerised datacentres do mean that they can add capacity fairly rapidly when needed though...

      1. Ken Moorhouse Silver badge

        Re: a lot of people decide to migrate a few thousand VMs on the same day

        The statement from which the above phrase was plucked should be on every single proposal to "The Board", putting forward the argument for not adopting a Cloud first policy.

        There are mechanisms in place to help prevent a run on a banking institution, but what would happen if everyone wanted access to their data in The Cloud at the exact same moment in time? Let's say you run a media organisation and some major event occurs. Would normal business for you be possible? Take 7/7 for example, the whole of the cellular network in london seemed to suffer an outage as presumably the emergency services had first dibs on bandwidth. If your organisation were required in some way to cover that event, you wouldn't be able to, even if you did have some eye-watering SLA's in place. Your average SME is sure to be affected - as mentioned in the original article, a company shutting down and starting up their phone system every day: that to me is a major risk factor, regardless of whether the idea is a prudent one or not.

        The government's reaction to any lobbying for action in the wake of such events would mean regulating cloud services, which would straight away give such organisations an excuse to hike prices. And guess, what, since you've long since disposed of your on-prem infrastructure, you've no option but to swallow those increases.

  11. Ken Moorhouse Silver badge

    We are very busy at the moment...

    It may take customers in excess of 25 minutes for their VM to be spun up. You may like to try again later or, if you have on-premises kit, use that instead.

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

Biting the hand that feeds IT © 1998–2019