buchachon - Fotolia

Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Why Microsoft Azure Stack is destined to fail

Microsoft touts Azure Stack as a way to bring public cloud benefits to on-premises environments. But, for many businesses, the sales pitch doesn't match their needs.

Microsoft Azure Stack promises to bring the managed, self-service nature of Azure's public cloud down to hardware that your organization owns and controls. Microsoft believes its hybrid cloud strategy requires this appliance, which fits between racked servers and Azure's giant server farms with a quintillion Xeon CPUs.

There's been a lot of talk about Microsoft Azure Stack, but it's going to fail for a few reasons.

Changes have plagued the project

It doesn't bode well that Microsoft Azure Stack has already been scrapped and restarted as something entirely different. There is an expression that no plan, no matter how well conceived, survives initial contact with the enemy. But it seems Microsoft was quick to rework Azure Stack when it encountered resistance to its first incarnation.

Then, in the latter half of 2016, customers complained when Microsoft noted Azure Stack would require new, specialized hardware from major vendors, such as Dell EMC and Hewlett Packard Enterprise. Furthermore, organizations that bought the new hardware would have to wait until mid-2017 to merge Azure with new on-premises hybrid cloud investments. Microsoft relented somewhat after significant negative customer feedback when it offered a "single-server" version of Azure Stack to do some of the tasking and tooling.

Between the product being continually reshaped and the limited availability of expensive Azure Stack-certified hardware, it's clear that a move to a Microsoft-designed hybrid cloud environment will be an expensive proposition -- and one that involves lifting and shifting, not transforming.

Self-service usually involves IT

The concept of replicating self-service automation and cloud-style control panels within an organization is misguided. In conversations with administrators and other IT personnel, what they want is either the public cloud for some workloads for features that public cloud providers uniquely offer or in-house control of the data so that they can manage the entire process from the start.

I have not found many IT departments that say the need to automate self-service with hybrid cloud on their data center hardware is a pain point. It's too difficult to develop this control plane on the disparate hardware and networking investments these departments have made over the years. Also, most business-side users lack the skills to use a hybrid cloud correctly, such as how to use a self-service private cloud panel, make intelligent choices about provisioning virtual resources for workloads, and make use of something that differs from available templates that require IT intervention.

Is hybrid cloud a solution in search of a problem?

I don't completely dismiss the hybrid cloud concept. There is plenty of room in the marketplace for options that work both on premises and in public clouds. Many organizations have business needs and software requirements that dictate certain workloads belong in the cloud for better fault tolerance, geographic distribution, bursting capability and so on. Many businesses will need to run applications and workloads in Microsoft Azure or Amazon Web Services (AWS), and they will need the tooling to manage those workloads, understand usage, monitor the operation and troubleshoot the failures.

Likewise, there are plenty of workloads that, no matter how hard Google, Microsoft and AWS try, will never run in the public cloud. Some businesses won't do it for regulatory reasons, or they don't have the technical expertise, or some more nuanced combination of the two.

It's time for Microsoft to realize the cloud is not the answer for everyone. For midsize organizations, the cloud is a limited destination. There are many reasons to do things inside your own data center. Microsoft is right about its concept of a hybrid cloud arrangement in which IT manages on-premises workloads and other workloads move into Azure or AWS as needed. However, it misses the bigger picture: Microsoft Azure Stack only addresses the need for self-service of in-house compute resources. It is a poor answer for a problem that many organizations just don't need to solve.

Next Steps

Take a closer look at Azure cloud adoption

Manage your Azure subscriptions

Find out which Azure automation tools are right for you

Dig Deeper on Microsoft Azure cloud services

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

What are the benefits and drawbacks of Microsoft Azure Stack?
Could not find a more idiotic and clueless person talking about Azure Stack, without even knowing what it is capable of doing, and this idiot must be a AWS employee who is feeling that his company is caught flat footed with a stupid architecture for a public cloud that they don't know how to respond to Azure
GlenBock, perhaps you can provide us with a more substantive response that goes beyond name calling?
GlenBock, perhaps you could enlighten us and offer a more substantive response to some of Jon's points?
I don't think this article does much to help anyone understand what the Azure Stack is and the title is at best inflammatory, which was probably the intention!? Whether you like Microsoft or not this is a significant step being taken here and I would prefer salient points for and against and use cases for all types of businesses, e.g. Enterprise, Service Providers, SMB's, Government, etc.
The first sentence in the final paragraph in particular doesn't make sense: "It's time for Microsoft to realize the cloud is not the answer for everyone." ... well that's one of the points of Azure Stack isn't it - to let customers run workloads in their own on premise datacentres? Typically to design and deploy an in-house platform with IaaS, PaaS and automation features is expensive and time consuming, and I would argue that to be able to purchase a fully featured cloud system will be an attractive proposition for a lot of companies. I'll expect Amazon and Google are already working on ways of bringing their Cloud platforms to customers and as the saying goes "Imitation is the sincerest form of flattery ..."
The idea is not new, time will show how business will adopt it
It seems to me that there is a need - how big is probably the question. This can be seen with Eucalyptus.
Having a similar on-premises platform that you can do some test coding against without the risk of your azure bill going crazy has merit as well.
And having the same management can also help - if Microsoft can keep the on-prem and online version in sync and managing everything with ie. PowerShell and the same cmdlets is also a boon.
Adding all the 3rd party powershell support really helps here.
There is an assumption that the self service is for end Users and hence, a training burden to doing it correctly.

An alternative view point is that it is abstracting away from the underlying physical infrastructure and the self service portal is more for the IT Admins so they don't have to worry about the underlying networking, etc.

I also think that the whole 'there might be a use case for unifying the way new capabilities are deployed across a hybrid estate' is a little understated.  My understanding is that TP3 will remove the requirement for phone home (which is a pre-requisite of billing based on usage) and instead make it optional - this in my opinion makes it much more usable across a larger number of use cases.

@JonHassell, Azure Stack currently only implements some of the features in Azure.  Some are not practical (e.g. big analytics that require 1000s of VMs) and some simply haven't been done yet.  One problem is that Azure Stack has to be API compatible with Azure - otherwise the whole hybrid thing doesn't work.  This therefore will likely mean that changes are slower than otherwise expected/wanted, especially if it requires changing Azure.  The general benefits are those of having a hybrid cloud strategy but currently it is tempered by limited feature set compared to Azure and the specific hardware requirement.  Over time you would expect these to be addressed but you would need to speak to MS about timescales.  Best bet is actually to download the Tech Preview and have a try - there is a one box install option for evaluation.  I personally see benefit if you are moving towards an automated infrastructure and wish to remove some of the complexity away from the integration.  I don't think it's quite there yet but it also hasn't been released yet :-)
Interesting read.
Biggest gripe with this, is not that Azurestack will fail, it may well do, but what is unfair is to suggest Microsoft don't understand the problem.

I'd suggest if the author had spoken to Microsoft, they would see that there is a clear vision for this solution and the problem they are trying to fix.

I spoke with them a month or so ago ahead of the TP3 launch (have a podcast on it if interested) and they understand the problem and more importantly their target market.

I think they fully understand that Azurestack is a specific use case, certainly not a general enterprise business solution and aimed much more at service providers than enterprises.

For a service provide a ready made Azure type solution, that you can roll into your data centre and have supported by major hardware vendors has the potential to be a very useful solution. And for those who want more certainty and a better support ecosystem this could be a very attractive alternative to OpenStack.

I think to look at Azurestack as some kind of bridge to the cloud for the average business, is to completely misunderstand the aim Microsoft have for it and fundamentally what a solution like this is designed for.

But heah, click bait is king - Azurestack is doomed :-)

Dan Coffin's response in here is excellent buy the way.. addresses both those points really well.
This is interesting to know that someone could predict the failure of one of the largest cloud provider.
The part I am struggling with is Microsoft wants me to buy $300,000-$400,000 worth of my own Hardware to make my own cloud (and using my own software licenses), so I can THEN pay Microsoft "transaction fees" to run my apps on my own machines with my own licenses. Ummmmmmmmmmmmmmm. Only advantage I see to this scenario is that the license fee's are less than running it on Microsoft's Azure servers. What's the return on spending $400,000 on my own servers?
Privacy and control. :p
Just calculate the 3 years billing on Azure cloud for the same resources capacity on your Azure Stack cloud you will find that you will save $200,000, With various other free services which is offered by Azure stack. Eg. Compliance certification, Multi Tenancy, Private cloud, Ease to manage IT and many more..