Introduction to Windows Azure Virtual Machines

Editor's Notes

• #5: Slide Objectives:Explain the differences and relationship between IaaS, PaaS, and SaaS in more detail.Speaking Points:Here’s another way to look at the cloud services taxonomy and how this taxonomy maps to the components in an IT infrastructure. Packaged SoftwareWith packaged software a customer would be responsible for managing the entire stack – ranging from the network connectivity to the applications. IaaSWith Infrastructure as a Service, the lower levels of the stack are managed by a vendor. Some of these components can be provided by traditional hosters – in fact most of them have moved to having a virtualized offering. Very few actually provide an OSThe customer is still responsible for managing the OS through the Applications. For the developer, an obvious benefit with IaaS is that it frees the developer from many concerns when provisioning physical or virtual machines. This was one of the earliest and primary use cases for Amazon Web Services Elastic Cloud Compute (EC2). Developers were able to readily provision virtual machines (AMIs) on EC2, develop and test solutions and, often, run the results ‘in production’. The only requirement was a credit card to pay for the services.PaaSWith Platform as a Service, everything from the network connectivity through the runtime is provided and managed by the platform vendor. The Windows Azure best fits in this category today. In fact because we don’t provide access to the underlying virtualization or operating system today, we’re often referred to as not providing IaaS.PaaS offerings further reduce the developer burden by additionally supporting the platform runtime and related application services. With PaaS, the developer can, almost immediately, begin creating the business logic for an application. Potentially, the increases in productivity are considerable and, because the hardware and operational aspects of the cloud platform are also managed by the cloud platform provider, applications can quickly be taken from an idea to reality very quickly.SaaSFinally, with SaaS, a vendor provides the application and abstracts you from all of the underlying components.
• #6: With the introduction of Windows Azure Virtual Machines and Virtual Networks there are an entirely new set of workloads that are suitable.
• #9: Cloud First Provisioning means exactly what is says. Building a VM in the cloud first. You have three methods of starting this process: Build a VM from the portal, from the command line OR programmatically calling the REST API. Once your choice of provisioning is made you will need to select the image and instance size to start from. The newly created disk will be stored in blob storage and your machine will boot.
• #12: This use case is when you already have a “golden image(s)” your company uses for server provisioning or you have a VM running on premises that you would just like to run in our data center. You take the vhd – use CSUpload to upload as a page blob to a storage account. From there use the portal to add as an image (sysprepped) or disk (configured VM) and there you can create a VM based off of the vhd.
• #13: This use case is all about using the capture feature of IaaS to create OS images. The first part you start with a base vhd that you then custom with software binaries, registry settings etc.. You run sysprep.exe and generalize/shutdown the OS. You can then upload it if coming from on-premises or just click the capture button if you created the VM in the cloud. Capture allows you to take a generalized VM and save the underlying VHD as a new image in your image library.
• #14: One of the key benefits of IaaS is flexibility and control. The Windows Azure solution provides the capability of not only moving VHDs TO the cloud but also allows you to copy the VHD back down and run it locally or on another cloud provider. Great for testing out production issues or any other need where you require a copy of the production server.
• #15: The OS and Data Disks are stored in Windows Azure storage. So in addition to the data being persistent you also get the benefits of storage which means your VHD is replicated 3X’s locally and also 3X’s in a separate data center in the same region (geo-replication)
• #16: This slide simply highlights that if the physical hardware backing your VM goes down a new server will start and pick up the same VHD.
• #21: This slide explains web and worker role architecture within a cloud service.Things to call out: Each role can have multiple identical instances You can currently have up to 25 roles per cloud service.
• #22: Virtual Machines work in a similar way.A cloud service can have up to 25 roles but if that role has a virtual machine there can only be a single instance. Implicit Cloud service is a reference to the portal. If you only have a single VM the portal will not display the cloud service although it is created and exists.Add another role (or remove the VM) and the cloud service will show.
• #23: When there are multiple VMs in the same cloud service they can communicate directly as they are on the same network.
• #25: In Windows Azure previously you did not have control over the VM names. This is a net new. Windows Azure Provided DNS is also new. Previously you had to use the Windows Azure object model to get the IP addresses of different roles in your deployment. BYOD allows you to host Active Directory directly in the cloud or when used in conjunction with Virtual Networks (VPN) to build hybrid solutions where AD is hosted on premises and consumed in the cloud or both.
• #27: Input port forwarding is how you can configure direct/non load balanced communication to multiple VMs that have services listening on the same port.
• #28: UDP is now a supported protocol in WA (previously wasn’t). Discuss that by default with virtual machines there is an internal endpoint defined with protocol=“all” so essentially all traffic is open between virtual machines. Windows Server firewall and Linux firewall will still need to be configured to allow traffic.Health probes allow the load balancer to check a custom HTTP path that tells the LB whether the server is healthy or not. For TCP endpoints no http path is required just the port to connect to.
• #30: Load Balanced Input Endpoints allow you to configure the load balance to distribute traffic to multiple VMs. This works with TCP/UDP and you can also configure an HTTP probe with a custom path to test the health of the VM.
• #31: Load Balanced Input Endpoints allow you to configure the load balance to distribute traffic to multiple VMs. This works with TCP/UDP and you can also configure an HTTP probe with a custom path to test the health of the VM.
• #34: Slide ObjectiveDiscuss achieving very high availability with Windows AzureSpeaker NotesFault Domain is a physical unit of failure, and is closely related to the physical infrastructure in the data centers. In Windows Azure the rack can be considered a fault domain. However there is no 1:1 mapping between fault domain and rack. Windows Azure Fabric is responsible to deploy the instances of your application in different fault domains. Obviously this only applies if you run more than 1 instanceRight now Fabric makes sure that your application uses at least 2 (two) fault domains, however depending on capacity and VM availability it may happen that it is spread across more than that. As a developer u have no direct control over how many fault domains your application will use but the way you configure it may impact your availabilityUpgrade Domain is a logical unit, which determines how particular service will be upgraded. The default number of upgrade domains that are configured for your application is 5 (five). You can control how many upgrade domains your application will use through the upgradeDomain configuration setting in your service definition file (CSDEF). Windows Azure Fabric ensures that particular upgrade domain is not within single fault domainThat is Upgrade domains are orthogonal to fault domainsRoles are then spread out over upgrade domains and fault domains. Subject to running a reasonable number of instances this removes the risk of a single point of failure
• #35: Availability sets tell the Fabric Controller to place VMs in the same set on different racks for faults and in separate upgrade domains for updates. This essentially tells the FC not to take the guest OS down of all VMs in the same set for host updates.
• #37: Explain that each tier of an application can be enabled with its own availability set which ensures at a physical hardware level in the data center that there is no single point of failure.
• #38: Without at least two virtual machines performing the same workload grouped into an availability set you only get the 99.9% SLA. To achieve 99.95 you must use availability sets.
• #39: We have now offer an SLA for virtual machines running a single instance. This is different from web/worker roles which require two VMs for an SLA.
• #41: Microsoft think about the stack to provide connectivity between on-premise and cloud.Specifically this deck focuses on the last two layersServicebusvs connect – SB requires app code change, Connect/Virtual Networks do not. Virtual Networks are the net new here. They provide site to site connectivity where Connect provided server to server connectivity.Virtual Networks are the more flexible and powerful option.
• #42: Windows Azure Virtual Networks is our solution to providing hybrid solutions and solutions that require advanced connectivity in the cloud. Hybrid on-premises to cloud is enabled by using the VPN solution that allows site<->site connectivity allowing machines on premise and machines in the cloud to appear on the same network.Advanced connectivity solutions are enabled because Windows Azure applications that are deployed in to a virtual network will have persistent IP addresses. This is a requirement for solutions like Active Directory. Other solutions enabled by virtual networks in the cloud are mixing VMs and web/worker role solutions in the same Windows Azure network. This allows for scenarios like web/worker roles to communicate back to VMs running applications like SQL server.
• #43: Windows Azure Virtual Networks provide functionality above and beyond site to site networking. VNETs also provide persistent IP addresses through an infinite DHCP lease (not static but effectively the same thing).VNETs also provide the capability of connecting cloud services in the same affinity group directly. This allows for scenarios like web and worker roles to virtual machines running services like SQL Server 2012 or SharePoint.