Google Compute Engine lets you run Windows virtual machines on Google's global infrastructure. Let's learn more about starting and running Windows VMs in GCE. We'll also discuss managing Windows deployments. In this module, we'll look at how to get up and running with Microsoft Windows Server and SQL Server on Google Cloud Platform using Google Compute Engine. Let's get started by looking at Google Compute Engine Fundamentals. Google Compute Engine is the infrastructures and service product on Google Cloud Platform. In this section, we'll look at the main features of the product and identify advantages of compute engine as a key foundation for your Windows workloads. When you get started with Google Compute Engine, you'll notice how easy it is to provision a virtual machine with the perfect configuration to suit your needs. You'll be able to select up to 64 virtual CPUs per machine, with between 0.9 and 6.5 gigabytes memory per CPU. You'll have the choice of standard, and SSD system disks, and local SSDs. You'll be able to get up and running quickly with available Windows Server images and Linux images. Well, you might be surprised to see Linux images in a class about running Windows workloads on GCP. Later on, you'll see how you can take your existing C-Sharp and dot net skills and apply them to code applications that run on Linux. Here's how you create a compute engine virtual machine with the cloud console web browser user interface. Should enter a suitable name for the VM and set the appropriate zone which will determine the Google data center where the machine will be created. Also, you can see the machine time that shows two virtual CPUs and 7.5 gigabytes memory, and the 50 gigabyte big disk with the Windows Server 2016 image. There are lots of other configuration options, but this shows all the basic choices that you will need to make. Once your machines are configured, you'll connect them to each other using the Google virtual private cloud network resource. You'll automatically benefit from Google's extremely high performance network, meaning that you get 2 gigabits per second of bandwidth per core between machines running in the same zone, with an upper limit of 16 gigabits per second of bandwidth. In addition, when you provision resources in different Google regions, traffic between them will flow exclusively over Google's own fiber network resulting in lower latency and higher throughput. Then if the traffic was routed via the public internet. To protect the virtual machines, there's a powerful, flexible, yet simple to configure firewall. You'll be able to make your Google resources available as a seamless extension of your on-premises environment using Google's virtual private network. Finally, we'll see later that Google's infrastructure will enable you to run your application with high availability and scalability taken care of with an HTTP and HTTPS load balancer, or the TCP and UDP load balance. At the time this slide was presented, Google's Windows Server images include: Windows Server 2008 R2 Desktop, and 2012 R2, and 2016 with both desktop and core versions. The price for the virtual machines provisioned on Google Compute Engine includes the Windows Server license. This costs four cents per core, per hour. This is pro rated permanent with a minimum of 10 minutes just like compute engines on hardware available. Next, let's look at Microsoft's SQL Servers report. Google supplies pre-configured compute engine images, the Express, Web, Standard, and Enterprise editions. With 2012, 2014, and 2016 versions available for all except text browse. Each SQL Server edition can be deployed on a variety of versions of Windows Server. Take a look at the docs for more information on this. Once again, the default price includes both the Microsoft Windows Server and SQL Server license cost, with the same permanent billing. However, in this case, you'll be able to bring your own license. If you need pre-configured Windows infrastructure that's ready to go, then you want to look at Google Cloud Launcher. This example shows the ASP.NET framework solution that comes with pre-installed Internet Information Services and ASP.NET. There's also a high availability SQL Server solution which will install multiple servers and configure the network to enable SQL Server always on availability groups. We'll take a look at that later on this module. Once your virtual machines are up and running, you want to connect them to perform General Systems Operations tasks, such as installing and configuring Windows features and your own applications. You'd be able to connect using Remote Desktop Protocol. And you can even do this without leaving your browser, by installing the current RDP for Google Cloud Platform extension. Of course you can also use Microsoft's RDP clouds instead. It's common to configure your Google Compute Engine Windows Virtual Machines using startup scripts. Startup scripts connect over on once on the first boot, in system preparation, on every boot, and can be written as batch, command, or partial scripts. This example shows a partial script that could be used to automate connecting the Windows compute engine instance to an active directory domain. Google compute engine offers very high quality infrastructure. It also offers extremely cost effective infrastructure. Here are four ways to save money on virtual machines. First, we've mentioned already that Google Compute Engine VMs are build per minute rather than per hour. Second, well there are a wide range of preconfigured virtual machines with up to 64 cores and 204 gigabytes of RAM, it's possible to select the perfect number of cores divisible by two in perfect amount of memory between 0.9 gigabytes and 6.5 gigabytes of RAM per core. For example, you can select the virtual machine with 18 cores and 36 gigabytes of RAM, if that's the best fit for your workload. Third, if you use standard VMs for a sustained period, you'll benefit from lower permanent cost. For example, if you run a machine for a complete month, then the per minute cost of that virtual machine will be 30% cheaper than the standard per minute charge. Fourth, if you have workloads that can tolerate interruption, such as rendering frames of video, you can run preemptible instances at 80% load per minute cost than standard. These machines have a maximum lifetime of 24 hours but may be preempted by Google. If the resources they consume are needed to run out the work.
