10 ways to optimize your Azure spend

The cloud holds an enormous amount of potential. Microsoft Azure, one of the most popular cloud computing platforms, offers organizations endless possibilities, no matter the industry or business challenge. Azure enables organizations to move and scale faster, lower IT costs and drive innovation, all of which are in demand now, more than ever.

However, optimizing cloud spend is key to managing costs and minimizing waste. A study by Flexera indicated that as much as 35 percent of public cloud resources are wasted. Areas of waste include over-provisioned or idle instances, abandoned cloud resources, and sub-optimal license assignment methodology. Very few organizations maximize their license assignment value, and most do not have a process to check compliance – often producing overspend. Organizations often struggle with a lack of visibility and insight into current costs and are challenged with controlling costs while mitigating risk.

If your organization is using Azure and is currently facing these challenges, read on to learn 10 ways you can optimize your Azure spend.

1. Right size compute and storage

Azure offers a wide range of virtual machines (VMs) for compute and storage. Each VM has a different capacity and different hardware and performance capabilities. If you are provisioning machines that offer more than what you need, then you’re paying for compute you don’t use. This isn’t all that uncommon. Many developers choose a VM with more than they need, especially if they’re worried about underestimating performance requirements.

By reviewing trends over time, you can determine if your team regularly provisions more compute and storage than needed and adjust estimates accordingly moving forward. Sometimes enabling auto-scaling measures allows for quick adjustment of compute whether the need grows or shrinks. This ensures that you only pay for what you actually need.

2. Utilize Spot VMs and B-Series VMs

Spot VMs are bid for based on available unused compute power on the Azure marketplace. Often this results in an 80-90 percent discount over standard pricing. This might be ideal for workloads that aren’t time-sensitive and can be stopped and started repeatedly.

B-Series VMs can be provisioned at a price 15-55 percent lower than standard VMs. This series is designed for applications that are typically idle but experience sudden usage bursts. These machines run with a low baseline amount of CPU power. When idle, your workload accumulates CPU credits. When there is a usage burst, accumulated credits go toward the additional capacity.

3. Clean up backup and compute instance

So-called "zombie assets" may be running on your VMs. These are programs or infrastructure components that aren’t needed and aren’t being used for anything. They’re simply taking up space and costing you money. Examples include VMs that were used for a particular purpose, then abandoned but not turned off, script errors that failed to deprovision a VM, idle load balancers, and more.

You can identify these “zombies” by searching for VMs with a Max CPU of less than 5 percent over the past month, as this is a common marker of such assets.

4. Optimize disaster recovery scenarios

There may be ways to better streamline your current disaster recovery strategy. A disaster recovery strategy is important to have in place as it can save not only money but corporate livelihoods in the event of a disaster. Azure Backup and Azure Site Recovery features allow for easy data and application backup to the cloud, as well as quick deploying of cloud-based replication, failover, and recovery.

You can also move disaster recovery content to Cold Locally Redundant Storage (LRS) to save storage costs. Keep in mind that the best practice is to move anything in the hot storage tier that hasn’t been used for a month to a cold storage tier. This can save 45 percent or more in monthly storage fees.

5. Monitor data transfer (cost) and optimize cross region peering

If your company has sites in varied locations or your employees are spread across the globe, you should take a close look at data transfer costs. When data transfers between Azure regions, you may be charged depending upon the zone of origination.

While two Azure networks connected through Azure VNET peering won’t incur data charges if they are in the same availability zone, all cross-zone traffic costs extra. Look for ways to minimize cross-zone traffic in order to reduce costs associated with cross-region peering

6. Leverage autoscaling and hybrid use benefits

Azure Autoscale is a built-in feature that allows applications to scale up and down rapidly to meet demand. Autoscale can be set to scale with different metrics, such as when memory use hits a certain threshold. You can also schedule autoscaling to meet predictable changes in demand.

By scaling up only when needed, you avoid paying for computing power that you aren’t using otherwise. You can also move on-premises workloads to the cloud when capacity needs arise and take advantage of the Azure Hybrid Benefit when doing so. This is a licensing benefit that lowers costs by letting you run workloads on the cloud if you’ve already paid for on-site licensing.

7. Migrate from database VMs to Elastic DBs

Using Azure VMs for database purposes can be very expensive. When it comes to VMs, database instances are often under-utilized, and load balancing is more complicated. Instead, you can save costs by moving to a Platform-as-a-Service (PaaS) model and transferring database workloads to Azure SQL Database services. The cost savings come from this model’s elasticity, making it so that you only pay for what you use.

8. Implement cloud-native features

Cloud-native features and applications are built specifically for the cloud. These tools are therefore optimized for the cloud environment. The underlying basis is a microservices architecture that enables various easily adaptable solutions built from managed services. If deployed strategically, this can also save money by improving productivity and increasing security and resilience.

9. Identify discounts for already-used services

any on-premises software licenses can be extended to cover cloud usage without requiring that you pay for a whole new set of licenses. Make sure you aren’t paying for additional licensing that you don’t need.

For example, if you are using Windows Server, Azure offers up to 82 percent cost savings. This total comes from combining Azure Hybrid Benefits, described above, with Azure Reserved Instances - an alternative to pay-as-you-go pricing in which VMs are reserved in advance.

10. Modernize workloads

Even if you need to keep some workloads on-premises to meet regulatory requirements or mitigate latency, or see a return on existing investments, you can still benefit from the Azure cloud by modernizing these workloads. Tools like Azure Stack HCI make it possible to consolidate on-premises virtualized applications onto a cloud-connected infrastructure so you can access the best price performance for storage and compute as well as increase your efficiency.

Stepping beyond native Microsoft Azure management features

Note that while the native Microsoft Azure Management features offer a great starting point when it comes to managing and tracking your cloud use, you may be able to achieve even more. Native Azure Cost Management tools include:

• Azure Cost Analysis: This provides a simple dashboard for visualizing your Azure cloud spend.
• Azure Budgets: This enables alerts via email or Azure Action Groups when budget thresholds are exceeded.
• Azure Advisor Recommendations: This offers cost-saving recommendations, including details about VM rightsizing and RI purchases.

However, independent cloud management tools may be able to provide insights missed by native applications. These tools allow you to maximize the cost savings of your entire software portfolio from a single platform. As your cloud consumption scales, Cloud management tools simplify the governance, budgeting, spend tracking, and chargeback of cloud costs to your organization’s business cost centers.