Azure Virtual Machines single instance SLA, a big step!

Hi all,

Early this week, Microsoft made an exciting announcement with its SLA for a single virtual machine of 99.9 ℅ : https://azure.microsoft.com/en-us/blog/announcing-4-tb-for-sap-hana-single-instance-sla-and-hybrid-use-benefit-images/

Before this announcement, single instance VMs (which are not part of an Availability Set) were not covered by an SLA. This was unattractive for workloads which do not support, afford or need a multi-instance deployment. This was generally applicable for legacy workloads and to be honest, to the majority of non-critical workloads, and for SMB workloads which do not afford investing in redundancy and HA.

Many of my customers avoided migrating workloads to Azure, just because of this, which was offered by AWS since a while.

With this announcement, customers have the warranty that an SLA of 99,9 % provided to their VMs, which means a maximum downtime of 8.76 hours per 356 days –> 44 minutes per month

Do not forget that this is only applicable for virtual machines with all disks stored on Premium Storage.

Optimize costs of your Azure Virtual Machines

Hi,

This is a very common topic which was discussed on many blog posts and forums. How to optimize the costs of the Azure virtual machines ?

There are many aspects of optimization, the first one which is also applicable for on-premises VMs is the size of the VM (Hardware optimization). If you track your Virtual Machine resource usage, you can see if your VMs are oversized, and then decide to resize them to reduce their costs. But this optimization, even if valuable, is a one-time optimization.

This is why I’m discussing here a better aspect of optimization, which is related to the uptime of the virtual machine.

The Stopped/Deallocated VM state

When you power off an Azure virtual machine, there are two possible states : Stopped and Stopped/Deallocated. The short story is that if you shutdown your VM to the Stopped state, your VM will be billable even if it’s actually stopped. But if you shutdown your VM to the Stopped/Deallocated state, your VM will not be charged during this downtime.

What is the difference, why not always use the Stop/deallocated mode ?

Many blog posts discuss the difference between both statuses. Here the key differences for ARM virtual machines:

– If you stop your Virtual Machine from inside the OS, or using a powershell/cli/api request without the Deallocate flag, the VM will be stopped but will not be de-provisioned from its Azure host (Hyper-V host). When you start your VM, it will start rapidly, and will keep all its network dynamic parameters (NIC IP Addresses)

– If you stop your virtual machine from the Azure portal, or using a powershell/cli/api request with the Deallocate flag, the VM will be sopped and de-provisioned from its Azure host (Compute resources). When you start your VM, Azure will redeploy the VM (the same VM Smile ) and you will notice a longer time for the VM to start. The VM will theoretically  have  different IP addresses if static IP addresses were not used.

What to optimize ?

You can, based on this property, schedule the Stop (Deallocated) and Start of your virtual machines that allow a downtime window during a period of time.

Examples

  • Test/Dev/Int/Rec virtual machines can be stopped during the week nights and during the weekends.
  • Virtual machines which are just used on a defined period of time can be stopped when not used
  • Even production virtual machines can be stopped if they are not used (A user file server can be stopped during the weekend)

Cost optimization gain example

This example is based on an existing SMB customer. This customer is planning to move all the non-production virtual machines to Azure IaaS. The first wave will include around 200 VMs, with the following  sizes repartition:

VM Count

Azure Size

Cost per month (744 hours)  (Windows based, NorthEurope)

Cost per month : VMs are stopped between 8PM and 7AM and Sunday ()

49

A1

€2,766.90

€1,305.35

56

D1v2

€4,286.50

€2,022.26

70

A2

€7,905.43

€3,729.58

26

D2v2

€3,980.32

€1,877.81

Total

€18,939.16

€8,935

Results:

  • A total gain of around 10K€/month –> 120k€/year
  • A cost reduction of around 55 %

How to implement it ?

There are many ways of achieving this goal, the more suitable is to use an Automation mechanism which Stop and Start the virtual machines based on a schedule.

The most suitable automation mechanism is Azure Automation, which can be used easily and without deploying any infrastructure.

There are many community participations to achieve this goal, but I like more the one published by Microsoft which give a more customizable downtime window per virtual machine, and using the ARM tags.

Here the link : https://docs.microsoft.com/en-us/azure/automation/automation-scenario-start-stop-vm-wjson-tags

What do we need to know about Azure Stack : The Q & A

Hi all,

It has been a long time since I didn’t blogged anything. A lot of news actually happened on the last few months, and in this post I will explain one of the most existing for me : Azure Stack

Azure Stack was introduced earlier this year (January) with a first Proof of Concept named TP1 (Technical preview 1). The Technical Preview goal was to give customers, consultants and early adopters a view of what Microsoft baptized as the future of Private and Hybrid cloud. But, really, what is Azure Stack ?

The modest definition 

Azure Stack is a platform (Software) that you can deploy on-premises to have similar Microsoft Azure services,  features and user experience. If you are using Microsoft Azure (The new portal, known as Ibiza portal portal.azure.com), than this is what you will get when you deploy Azure Stack on your datacenter. You will be able to leverage the Azure underlying technology on-premises, to deploy, manage and benefit from the Azure services like virtual machines, web apps, virtual networks and the list evolves. Think just that instead of typing portal.azure.com on your browser, you will type a custom URL on your domain that will land you on an Azure portal, but on your datacenter.

Is Azure Stack suitable for my company or business ?

Azure Stack brings advanced cloud technologies to your datacenter from the virtualization platform (a simple virtual machine) to the Azure App Services (A PaaS model to rapidly deploy Web Applications). So technically, Azure Stack can be used by any company aiming at least to use virtualization, but this is not enough to adopt it. As a consultant and  an Azure Architect, I think that Azure Stack is suitable for you if :

  • You are using or at least experimented the user experience, concept and different services provided by Microsoft Azure. If you have validated that Azure is suitable for your company, and you are looking for the same experience on-premises (for any reason) then Azure Stack may be a good choice (Azure Stack is consistent with Azure)
  • You are looking for a  private cloud platform which can provide the last cloud technologies and concepts. Azure Stack is born from Azure and will continually benefit from the last enhancements made and tested on the Azure public cloud platform
  • You are looking for a modern way to faster build your applications and services, which is the model based on PaaS and micro services. Azure Stack in its first version (mid 2017) will support Azure Web Apps and maybe Azure Fabric if they decide to bring it.
  • The constraints  I will mention next do not bother you

How Azure Stack will be delivered to customers ?

This is the actual debate, but Microsoft elected the winner, with a sort of inflexibility. Azure Stack will only be provided via System Integrated platforms with the freedom to choose between 3 different Hardware providers : HPE, DELL EMC and Lenovo (formerly x86 IBM servers). This means that you cannot deploy an Azure Stack platform on top of your hardware, but you will need to acquire the hardware with Azure Stack pre-packaged, and just plug it to your datacenter.

This last statement created a rage from the community, and we got two visions :

  • Microsoft is affirming that this model is the only possible way to achieve the wished Enterprise level private/hybrid cloud platform. Microsoft is stating that the integration with the hardware is a very heavy task and it prefers validating the platform and then provide a ‘ready engine’ to the customer.
  • The community is surprised that Microsoft is, first locking-out its solution for a set of non-affordable hardware providers, and secondly, not following the original virtualization and cloud ideology, which is the reuse and the optimization of the existing resources, and even, use affordable and commodity hardware.

Azure Stack licensing and prices

This is what I call the mystery question for the public because I have an early information, but due to NDA, I’m not allowed to publish it. What I can say, is that whatever the licensing model is, I think that it will be expensive, and I wonder if it will reach the SMB market. Remember, there are 3 parties involved : The hardware provider, the software provider and the integrator (which is Microsoft anyway but should be counted as a third party IMHO)

 

What if I acquire Azure Stack ? What about the test platform ?

This is a question I have asked before, and the answer was quite not sufficient. This is the summary :

  • Microsoft will provide the one Node PoC which is a one node Azure Stack platform. It’s what delivered today on TP1 and TP2. You can install Azure Stack on one node to be able to PoC, discover and make the tests you want. But, on the meanwhile we are not certain (no information) of the accuracy of the one node PoC with the System Integrated Azure Stack platform in terms of minor updates and bug fixes, and more important features.
  • You can do what you actually doing on Azure : Create a Test subscription with limited quotas, where you can make deployment tests  –> This still depends on the licensing model as we don’t want that a test platform be costly

 

What are the actual sizes of the Azure Stack Integrated System offers ?

It’s too early to speak about the capacity (CPU, Memory, Storage) of the Azure Stack platforms provided via the Integrated Systems. Things can change and I think that the final sizes will be revealed during the GA. Anyway the minimum published size today is  4 hosts with 256 GB and double sockets with 8 cores each constituting a single scale unit (Hyper-V cluster) and which will contain both the management infrastructure (VMs needed for Azure Stack to work) and your workloads (IaaS VMs, App Service Plan VMs…). Do not forget that the Azure consistency implies that the virtual machine sizes that you will be able to deploy are the same than Azure VM sizes. Hans Vredevoort has thorough articles about the Azure Stack Architecture :

Where System Integrators are placed on this whole thing ?

This is one of the questions I asked to myself. On standard products  like System Center and Windows Azure Pack, system integrators were almost mandatory to successfully deploy these products within a customer site. But the question raised up with the decision to only provide Azure Stack via System Integrated : No more need for system integrators to deploy the solution, a sort of plug and play.

This is true, but this isn’t so bad (No the case  for geeks, unfortunately Smile )

In fact, what are you integrators doing today with your customers when they call you to help them design and deploy their workloads on Azure? you are certainly happy, and this is why we should be optimistic when speaking about Azure Stack.  Because Azure Stack is Azure on your datacenter (An Azure sealed black box), customers will need you to help them first choose which Azure Stack offer (tier) to purchase, and then help them use Azure Stack, the same what you are doing  today on Azure. The consistency will make your Azure expertise vey valuable on the on-premise field.

What we will miss, is having our hands on an Azure Stack real platform to make some practice. But, theoretically, this will not be the biggest problem since we can use the one node PoC to achieve such goals. What is causing the headache for the com munity, is the near-impossibility of deploying the 1 Node PoC on our LAB at home. The minimum RAM requirement for the TP2 is 96 GB of RAM, and this is the minimum, expect up to 128 GB to start enjoying the LAB and the deployment of all the services. I don’ t know many having a real 128 GB RAM server at home.

 

Can CSP benefit from Azure Stack ?

Things are not yet dry, and Microsoft did not yet publish a clear view of the Cloud Service provider interaction with Azure Stack. But, it’s clear that that through the CSP program, CSP will be able to use Azure Stack to deliver sophisticated Azure like features. The biggest factors that may slow down CSP form using Azure Stack are :

  • Locked-down hardware providers : Cloud Service Providers are certainly partnering with hardware providers to acquire discounts and advantages when buying hardware. I’m very pessimistic regarding this factor. CSP may look to other cloud platform solution or continue build their own
  • Licensing and pricing : The introduction of  locked-hardware model may impact the margin the CSP can generate. No comment on the Software part licensing.

 

What do I think of Azure Stack and the implementation model ?

Microsoft is a great Enterprise, master minds are working there trying every day to enhance their products, creating new technologies and approaches, pivoting and changing their business model to impact the market. But, no human is bullet-proof, Microsoft can make mistakes and failures (The case for Windows phone, which is terribly not progressing ), it’s dramatically changing its business model with Azure Stack : Cloud Appliances. I’m really waiting for the licensing announcement to see which customer segment it’s targeting. But what I’m waiting for is the customer reaction. I have no idea how they will react. My first impression is that this model is not appreciated by the community and by me. I’m not against System Integrated platforms, but I’m with the virtualization and cloud early goals : Hardware reuse and cost reduction. Azure Stack is not fitting there, in addition of making restrictions on the hardware providers we can choose from, I have a bad feeling melted with a great excitement to this Azure Stack era, hoping it will find its way.

Based on my opinion, the following are the Pro and Cons of Azure Stack :

The GOOD

    • Azure services on your datacenter : This is the most exiting about Azure Stack, you can bring Azure features (IaaS, PaaS…) on your datacenter. You have the last tested*  cloud technologies on your hand. If you are avoiding using the public cloud platforms for any reason (Privacy, Compliance, trust, network connectivity) but on the same time wishing using the provided features, then Azure Stack is for you
    • Plug-and-Play model : The System Integrated model will reduce the TCO, by bringing a ready-to-use private cloud platform
    • A true Consistent and Hybrid cloud platform : Azure Stack is a real advantage for customers using or planning to use Azure services, since the consistency between the platform is guaranteed. You can use the same approaches, design decision factors, tools, scripts and features. You no longer need two different models to manage your cloud and on-premise platform, thus reducing significantly the IT efforts which can be spent on real business concerns (Deploying apps, migrating, enhancing…)

The Bad

    • The other side of System Integrated model : Hardware locking was never a good idea, depriving the customer from freely choosing  its hardware provider, and hence better control the costs. This model will reduce the early adopters market and hence can slow down this beautiful product from being used widely.

* Azure Stack will bring features already used on Azure, so we are sure that the features were widely tested on the public cloud

If you think the System Integrated model is not suitable for you, support this idea, you can participate and influence changing the Azure Stack future : Provide an installable version of Azure Stack