Blog post

Further Q&A from AWS vs Azure Webinar

By Kyle Hilgendorf | September 25, 2014 | 2 Comments


Last week I conducted a Gartner webinar comparing Amazon Web Services against Microsoft Azure.  You can watch the replay here.

There were an incredible number of questions that came in electronically during the webinar and I was unable to get to all of them in time.  Here are some additional questions submitted with a brief answer from me.  I have purposely not edited the submitted questions.  These are not all of the questions.  If I have time I will address more in a second blog post.

Finally, several people asked why they could not access Gartner’s “Evaluation Criteria for Cloud IaaS Providers” or the In-Depth Assessments for AWS or Azure.  Gartner has several subscription models and these documents are only accessible to Gartner clients that have access to “Gartner for Technical Professionals”.  If you do not have access and would like to discuss this, please contact your Gartner Account Executive.


Q: Isn’t it more realistic to think that we’re likley to end up using BOTH AWS and Azure due to the pervasiveness of Microsoft across the enterprise (Exchange, Sharepoint) even in shops that are not .NET shops?

I think it’s very possible that large organizations will find themselves using multiple cloud providers at the same layer (e.g. IaaS, PaaS) in the future.  Today, only a minority of my clients are using AWS and Auzre simultaneously but I think the rationale for this will increase over time for most organizations.  It might not be AWS and Azure but I see a day were large organization prefer to have 2-4 major IaaS/PaaS providers in place in order to deploy workloads to a best of breed scenario.  If your organization heads down this path, please do not underestimate the work involved in managing multiple provider relationships nor the effort involved in managing assets simultaneously at multiple providers.  This will impact your processes, your people and your integration points.

Q: Linux Virtual machines / Linux platform support on Azure?

 What is your feel on Azure’s commitment to Linux as a platform in general , 
I see very little technical documentation, videos, teched/ azure friday, Microsoft virtual academy resources on Linux support in azure.
 Every thing i see read and experiment is windows and windows only 
Given that how can we even roll some thing enterprise class based on a linux platform with no hand holding committment from Microsoft Azure ?

First and foremost, I do not represent any vendor, including Microsoft.  With that being said, every indication to me from Microsoft has been that they are committed to supporting both Linux and open source workloads.  Azure already supports a variety of Linux distributions as well as SUSE Linux Enterprise Server.  A relationship has not been worked out with Red Hat yet for RHEL, and I understand that can be a sore spot for many organizations.  The question I can’t answer is the degree of ease and automation that Microsoft will offer for non-Microsoft workloads.  We already know that Microsoft has put forth great effort to automate and orchestrate the provisioning of complex Microsoft stacks atop Azure like SharePoint through PowerShell scripts and cmdlets.  Time will tell as to whether Microsoft will do this for Linux or whether a community of Azure users will take this upon themselves.  Every indication I have from Microsoft however is that Microsoft will continue to push forward with more non-Microsoft software and platform support, especially Linux.  I encourage you to discuss this with your Microsoft account manager and ask for a private conversation with Azure leadership.

Q: Is data sovereignty an issue that you come across and how do these providers deal with that?

Data sovereignty is an issue that comes up in client conversations, but not as much as it used to.  In the webinar I talked about Local and Global availability and the differences between AWS and Azure.  Both providers have protections in place whereby you as the customer can be assured that data does not cross country or geographic boundaries.  AWS gets a slight advantage here because if you want to increase availability at certain levels with Azure you have to setup datacenters pairs – potentially across country boundaries (e.g. Ireland + Amsterdam or Brazil + U.S.).  However, at the same time, you can choose not to do this with Azure if you prefer to keep your data within a single location.  Data sovereignty issues still come up for clients that have specific country requirements like Germany, France or Canada – locations where these providers do not yet have datacenters.

Q: Does the AWS custom tagging allow you to create multiple groups of assets using different permutations of tags i.e a server can live in multiple overlapping groups e.g. a billing group vs an application group vs a project group

Good question.  You can assign multiple tags to AWS assets.  For example, you could setup tags for department number, billing group, project name, etc.  Furthermore, you can have each of these tags pass through into your bill and then use a custom reporting tool (most use Excel Pivot Tables) to sort and filter based on whatever combination of tags you prefer.  Check out this documentation:

Q: How do you compare the storage offerings from Azure to AWS.
The non-existent Virtual block storage as an offering in Azure is a big issue to run realworld database work loads.
The fact that BLOBs, Queues, Tables are not BLOCK storage based but Object storage based is huge problem for any enterprise workloads.
Currently Azure supports only 500 IOPS at 512-2K chunk size
They are working on a PIOPS model but the IOPS is not a enterprise level.

What are your views on this ?

My colleague Angelina Troy is the resident expert on this.  She has published an in depth comparison of public cloud storage services from AWS, Azure and Google.  Gartner for Technical Professional clients can access that document here.

Q:I have internal requirements from Federal Government, What provider can be a better option for a Government?

This is highly dependent upon your specific federal requirements – they vary quite a bit.  AWS does offer GovCloud, which is a unique ITAR-compliant region just for agencies or contractors that meet certain specifications.  Microsoft does not yet have a government-only region of Azure in general availability.  We do know of some government entities using Azure though despite this.  If you are a federal agency and interested in a federal cloud from Azure, I encourage you to discuss the options with your Microsoft representative for private preview/beta.

Q: Does azure have the same concepts of pay-by-the -minute, spot or bid pricing?

Azure does offer per-minute billing but does not have an auction style model like AWS’ spot pricing.

Q: Regarding: AWS can scale applications on demand.

Can this feature be used to lower the pricing during idle times by a lot?

I’d need more specific context about the application design – but in theory the answer is yes.  Let’s say that your application is horizontally scaled out to 10 web servers during the day to handle load.  If you setup Auto Scaling, Elastic Load Balancing and CloudWatch monitoring appropriately, you could start to scale those 10 web servers down to 8, 6, 5, 2 or 1 web server during the evenings/weekends and then back up when you need them.  Considering that at least 6-8 hours of each day could be a trough for your application, you might be able to realize up to 30-40% savings over a static design.  There are a lot of intricacies into how you setup such an application, but there is plenty in the industry published about “cloud-native” designs or “scale-out” and “scale-in” designs.

Q: I have not heard good things about Microsoft support, even when a customer has Premier Support. Is Amazon even worse?

When you have the number of client conversations over a 3+ year period that I’ve had, you hear many positive and negative comments on just about every vendor’s support.  With that said, I have heard mostly positive support on both Azure and AWS support plans.  The one benefit to Premier Support that I highlighted in the webinar is that Microsoft Premier Support will not stop at only helping with “cloud issues”.  If you are running Windows Server and a .NET application atop Azure, Premier Support will assist you with Windows Server or .NET issues as well as Azure issues.  That is a nice touch for a complete Microsoft stack.  AWS will offer best effort support for things like Windows Server – but they are not Microsoft – they can’t be expected to provide intimate support for Windows Server.  With all this said, given enough time you will probably eventually have a bad experience with every support organization.  Any of us that have called our home broadband or cable/satellite TV support could attest to this!  However, I’ve never sensed a trend that says Azure or AWS support is bad.

Q: Whats a strong platform that integrates well with Azure/AWS that enables development in Azure/AWS, while abstracting out the dependencies, providing portability?

Great question.  This could mean several things so I may have to guess at the intent to the question.  It sounds like you are interested in adopting an abstracted development platform that allows you to develop against either AWS or Azure and port back and forth as needed.  This idea is mostly a fantasy utopia at the present time, but certain things are intriguing.  If you are interested in abstracted APIs that translate to multiple cloud providers you might want to look at Apache jclouds, Apache Libcloud or Dasein.  Just pay careful attention about the support for or lack thereof of various providers.  If you are looking for a broker that helps you manage both providers through the same process, check out RightScale, CSC Agility Platform or Dell Cloud Manager.  But we have to remember that these are very different services architected differently.  If you want to deploy a database schema to AWS RDS and then port that to Azure SQL – you’ll need to do some manual work.  If you want to deploy to Azure Load Balanacing (ALB) and then move to AWS Elastic Load Balancing (ELB) – you’ll find nuances and differences between the designs.  Therefore, portability is far from a reality and I doubt we’ll ever really see it.  A day where these services are identical means that differentiation has died.  And competitive capitalism will not let that happen.

My advice to clients interested in this is to avoid proprietary features whenever possible.  For example, instead of choosing DynamoDB or Azure Tables for a NoSQL database, rather opt for a virtual machine with MongoDB, Redis, or CouchDB atop of it.  In that scenario, you can always redeploy a VM with the other provider, reinstall the database and migrate the data.  Or rather than ALB or ELB for load balancing consider HAProxy or a virtual appliance load balancer that is supported at multiple providers.  However, provider’s proprietary services are popular because they are cheap and easy to integrate with other services at that provider.  So that tug and pull will aways be there.

Q: Which provider is more flexible in terms providing scalability? AWS has certain restrictions like no vertical scalability , can’t increase individual components like vCPU or RAM of an instance , also block storage volume can’t be extended beyond 1 TB. What are your inputs on this ?

My webinar highlighted that AWS is the choice if you need the highest levels of scalability.  That is not to disparage Azure, its simply supported by multiple data points from our Evaluation Criteria research.  AWS can increase individual components like vCPU and RAM in a vertical scalability concept – its simply a bit different than their horizontal scalability.  You can shut down an EC2 instance and resize it (e.g. M3.medium to M3.large) as long as the processor architecture and OS type (e.g. 64-bit) is the same.  I find a lot of AWS customers don’t know this, but it is possible to achieve.  Just be careful when moving from micro or small because those often are 32bit or single processor architectures.

It is true that EBS has a 1 TB limit.  However, Azure’s block blob size limit is currently 200GB.  So if you are in need of block volumes larger than 1TB, I think you are out of luck right now with either provider unless you contact the provider and ask if something special can be arranged for you (sometimes it can).

The Gartner Blog Network provides an opportunity for Gartner analysts to test ideas and move research forward. Because the content posted by Gartner analysts on this site does not undergo our standard editorial review, all comments or opinions expressed hereunder are those of the individual contributors and do not represent the views of Gartner, Inc. or its management.

Comments are closed


  • Kyle,
    Great vision on the complexities of end users and the perception of cloud portability/contestability….
    You painted a picture in my head of a future portal for cloud services which goes ‘Red’, in cloud portability, as you pick cloud specific options for your deployment.

    Your example focusing on Load Balancing and the optimum services offered by each cloud provider is ‘Spot On’.

    I see the reality of Development and Test as a Service (DTaaS) virtual assets as being the first DevOps area, which can realize the portability to ‘Any Cloud’: ‘Best Price’ portability,, with promotion to the downstream DevOps areas,..QA, Pre-Prod, Prod… being more and more restrictive as it relates to Hybrid Cloud Portability.


  • The usual rates for payday or payday loan range from 8-30 per 105 borrowed,
    per week.