As a home improvement retailer, the proverb measure twice, cut once is fitting for the team at Tailwind traders. Here are some recommended practices that can help you minimize your costs. First, it's important to understand estimated costs before you deploy. To help you plan your solution on Azure, carefully consider the products, services, and resources you need. Read the relevant documentation to understand how each of your choices is metered and billed. Calculate your projected costs by using the pricing calculator and the TCO calculator. Only add the products, services, and resources that you need for your solution. Ideally, you want your provisions, resources to match your actual usage. Azure advisor identifies unused or underutilized resources and recommends unused resources that you can remove. This information helps you configure your resources to match your actual workload. Azure advisor recommendations are sorted by impact, high, medium, or low. In some cases, Azure advisor can automatically remediate or fix the underlying problem. Other issues, such as the two that are listed as high-impact, require human intervention. If you have a free trial or a credit-based Azure subscription, you can use spending limits to prevent accidental overrun. For example, when you spend all the credit included with your Azure Free Account, Azure resources that you deployed are removed from production and your Azure Virtual machines are stopped and de-allocated. The data in your storage accounts is available as read only. At this point, you can upgrade your free trial subscription to a pay-as-you-go subscription. If you have a credit-based subscription and you reach your configured spending limit, Azure suspends your subscription until a new billing period begins. A related concept is quotas or limits on the number of similar resources that you can provision within your subscription. For example, you can allocate up to 25,000 VMs per region. These limits mainly help Microsoft plans its data-centric capacity. Azure reservations offer discounted prices on certain Azure services. Azure reservations can save you up to 72 percent as compared to pay as you go prices. To receive a discount, you reserve services and resources by paying in advance. For example, you can pre-pay for one year or three years of use of VMs, database compute capacity, database throughput, and other Azure resources. This example shows estimated savings on VMs. In this example, you save an estimated 72 percent by committing to a three-year term. Azure reservations are available to customers with an enterprise agreement, Cloud Solution Providers and pay as you go subscriptions. The cost of Azure products, services, and resources can vary across locations and regions. If possible, you should use them in those locations and regions where they cost less. But remember, some resources are metered and billed according to how much outgoing egress network bandwidth they consume. You should provision connected resources that are metered by bandwidth in the same Azure region to reduce egress traffic between them. Keep up to date with the latest Azure customer and subscription offers and switch to offers that provide the greatest cost-saving benefit. Azure cost management and billing is a free service that helps you understand your Azure bill, manage your account and subscriptions, monitor and control Azure spending, and optimize resource use. This image shows current usage broken down by service. In this example, Azure app service, a web application hosting service, generates the greatest cost. Azure cost management and billing provides a number of features. With reporting, you can use historical data to generate reports and forecast future usage and expenditure. Data enrichment allows you to improve accountability by categorizing resources with tags that correspond to real-world business and organizational units. You can create and manage cost and usage budgets by monitoring resource demand trends, consumption rates, and cost patterns. With alerts, you can set alerts based on your cost and usage budgets. You can receive recommendations to eliminate idle resources and to optimize the Azure resources you provision. Tags help you manage costs associated with the different groups of Azure products and resources. You can apply tags to groups of Azure resources to organize billing data. For example, if you run several VMs for different teams, you can use tags to categorize costs by department, such as human resources, marketing or finance, or by environment, such as test or production. Tags make it easier to identify groups that generate the biggest Azure costs, which can help you adjust your spending accordingly. This image shows a year's worth of usage broken down by tags on the Azure Cost Management Plus billing page. A common recommendation that you'll find from Azure Cost Management and billing and Azure advisor is to re-size or shut down VMs that are under-utilized or idle. As an example, say you have a VM whose sizes Standard D4_v4, a general purpose VM type with four V CPUs and 16 gigabytes of memory. You might discover that this VM is idle, 90 percent of the time. Virtual machine costs are linear and double for each size larger in the same series. In this case, if you reduce the VM size from Standard D4_v4 to standard D2_V4, which is the next size lower, you reduce your compute cost by 50 percent. Keep in mind, the resizing of VM requires it to be stopped, resized, and then restarted. This process might take a few minutes depending on how significant the size change is. Be sure to properly plan for an outage or shift your traffic to another instance while you perform resize operations. Recall that to de-allocate a VM means to no longer run the VM, but preserve the associated hard-disks and data in Azure. If you have VM workloads that are only used during certain periods, but you're running them every hour of every day, you're wasting money. These VMs are great candidates to shut down when not in use and start back when you need them, saving you compute costs while the VM is de-allocated. This approach is an excellent strategy for development and testing environments where the VMs are needed only during business hours. Azure even provides a way to automatically start and stop your VMs on a schedule. This recommendation might sound obvious, but if you aren't using a resource, you should shut it down. It's not uncommon to find non-production or proof-of-concept systems that are no longer needed following the completion of a project. Regularly review your environment and work to identify these systems. Shutting down these systems can have a dual benefit, by saving you an Infrastructure costs and potential savings on licensing and operating costs. As you move your workloads to the cloud, a natural evolution is to start with infrastructure as a service or IaaS services because they map more directly to concepts and operations you're already familiar with. Over time, one way to reduce costs is to gradually move, IaaS workloads to run on Platform as a Service or PaaS services. While you can think of IaaS, as direct access to compute infrastructure, PaaS provides ready-made development and deployment environments that are managed for you. As an example, say you run SQL Server on a VM running on Azure. This configuration requires you to manage the underlying operating system, setup a SQL Server license, manage software and security updates, and so on. You also pay for the VM, whether or not the database is processing queries. One way to potentially save costs is to move your database from SQL Server on a VM to Azure SQL database. Azure SQL Database is based on SQL Server. Not only are PaaS services such as Azure SQL database often less expensive to run, but because they're managed for you, you don't need to worry about software updates, security patches, or optimizing physical storage for read and write operations. Licensing is another area that can dramatically impact your cloud spending. Let's look at some ways that can reduce your licensing costs. You should also choose a cost-effective operating system. Many Azure services provide a choice of running on Windows or Linux. In some cases, the cost depends on which you choose. When you have a choice and your application doesn't depend on the underlying operating system. It's useful to compare pricing to see whether you can save money. If you've purchased licenses for Windows Server or SQL Server and your licenses are covered by software assurance, you might be able to re-purpose those licenses on VMs on Azure. Some of the details vary between Windows Server or SQL Server. We'll provide resources at the end of this module where you can learn more.
