Cloud computing provides a variable cost model and reduces the need to budget for on-premises data center spending. However, there are many budgeting pitfalls you might encounter that unintentionally drive up your cloud costs.
Platforms like Microsoft Azure give you plenty of tools to help control cloud costs, but it’s not always easy to know where to begin. Here are 12 tips to help you lower Azure spend and get more ROI from your Azure environment:
1. Understand how Azure pricing works
Whether you are making the jump from Amazon Web Services (AWS) to Azure, deploying a multi-cloud environment, or all-in on Microsoft’s cloud platform, you should have a clear understanding of how Azure’s pricing model works. The best way to lower Azure spend is to know what you’re spending money on, so getting familiar with their pricing structure is the best way to begin saving. Like AWS and other cloud providers, Azure offers a number of services across compute, storage, networking, and data analytics. If you’re new to Azure, that also means being aware of how the terminology differs from platform to platform.
Azure pricing covers a number of popular services and technologies, from compute and containers to storage and security. In many parts of the world, Azure will bill for services in U.S. dollars. Take note of any exchange rates or ask your cloud solutions provider for more details on your Azure pricing.
2. Take advantage of reserved VMs and discounts
Azure Reserved Virtual Machine (VM) Instances work like AWS Reserved Instances (RIs) in that they enable you to pay in advance for your virtual machines. Like AWS RIs, Azure RIs save about 72% on cloud computing costs when compared with pay-as-you-go pricing, according to Microsoft.
When used alongside Azure Hybrid Benefit, Azure RIs can save up to 85% compared to pay-as-you-go pricing. Azure Hybrid Benefit is a service for hybrid cloud users who want to leverage their on-prem Windows Server and SQL Server licenses in the Azure cloud.
3. Optimize workloads by VM type
Paying attention to which types of resources you’ve provisioned can help to significantly lower your Azure costs. Understanding which Azure VM types are ideal for the workloads you want to run is important when trying to lower Azure spend. Azure VMs are broken out into several different types based on their overall purpose. Similar to AWS EC2 instance types, Azure VM types and their series include:
- General Purpose: Bs- and D-Series (and previous generation A-Series) — Designed for testing, development, small and medium databases, and low to medium-traffic servers
- Compute Optimized: F-Series — High CPU-to-memory ratio, designed for medium-traffic web servers, network appliances, batch processes, and application servers
- Memory Optimized: E-, M- Mv-, and D-Series — High memory-to-CPU ratio, for relational database servers, medium to large caches, and in-memory analytics
- Storage Optimized: Ls-Series — Ideal for Big Data; data warehousing; and SQL, NoSQL, and large transactional databases
- GPU: N-Series — Specialized VMs for graphic rendering, video editing, model training and inferencing, deep learning, and remote visualization
- High Performance Compute: H-Series — The most powerful VMs currently available
4. Right size your VMs
Right sizing means making sure that what you have provisioned is neither too much nor too little capacity for what you need. Overprovisioning means that you’re paying far more than you need for VMs or other Azure resources. However, if you under-provision resources, then you run the risk of not being able to meet demand.
Using the list above, for example, if you’re running H-Series VMs for high performance (starting at $581 per month) when what you really need are D-Series VMs designed for general purpose computing (starting at about $42 per month), you could be spending hundreds of dollars extra per month than is necessary.
5. Move (your data) to another region
Azure costs vary by region, or the set of data centers within a specific geographic location. Each region then contains different availability zones, with their own independent power, network, and cooling to provide redundancy in the event of a data center failure.
Consider the geographic location of your resources when estimating costs. When possible, you can also elect to move workloads to another, less expensive region. Wondering where your data can or should live? Think about where your SaaS solutions live when thinking about your sovereignty, latency, security, and cost.
6. Shut down idle and unused VMs
Going from data center purchasing habits — buying the largest available computing capacity — to spending in the cloud can be a bit of a culture shock. While you do need to plan for peak traffic, the cloud is elastic in that you don’t have to run at full power 24/7.
Using a cloud management platform, you can check for idle, unused, or underutilized VMs to better utilize your cloud resources. Idle VMs are VMs that were once running but have been left on, driving up costs. Meanwhile, VMs that have been unused or underutilized were purchased but have not been taken advantage of.
With any cloud platform, you don’t pay for what you use — you pay for what you order. Clearing up resources that you no longer need can help significantly lower your cloud bill.