How to Reduce Azure Costs: Azure Cost Optimization
When moving workloads to the cloud, it’s easy to fall into old habits – like provisioning virtual servers without a holistic understanding of the technical requirements that will affect how the servers are utilized. Companies that are not actively managing the resource consumption for their Microsoft Azure subscriptions are often overpaying for oversize costs and over-provisioned services. Consequently, they are overspending on idle and underutilized resources.
Without an all-encompassing view of the context of the application, workload, or environment it is nearly impossible to avoid overpaying. By managing the resource consumption of cloud services, organizations can make well-informed decisions and lower the risk of unnecessary spending.
There are Azure cloud cost savings that can be gained by adjusting:
- Application Usage
- Storage Tiers
- Reserved and Burstable Instances
With an Azure virtual desktop, you get the CPU, disk, and up/down state of your server from the platform. Enabling Azure Monitor provides additional insights into the performance and dependencies of your virtual machines.
Perhaps a server that IT provisioned several months or even years ago no longer needs its level of resources and can be downgraded from a sizing perspective. Through Azure Monitor, performance metrics can be tracked and visualized to provide real-time and performance history analysis.
All data collected by Azure Monitor collects fits into one of two fundamental types: metrics, and logs. Metrics are numerical values that describe some aspect of a system at a point in time. They can tell you how your server is performing but typically need to be combined with logs to identify the root cause of issues. The IT team can use both sets of data in a Log Analytics workspace to then "tune" the virtual machine to its optimal level of performance.
Azure Monitor ultimately provides critical IT insights to recommend potential changes to the business from a service delivery standpoint. By right-sizing line-of-business applications to the Azure consumption necessary to run the application, the total cost of ownership (TCO) can be achieved.
Azure Advisor is a powerful tool built right into the Azure platform for all customers to analyze the consumption of their cloud services. Azure Advisor helps you optimize and reduce Azure costs by identifying idle and underutilized resources.
You can get cost recommendations from the cost tab on the Advisor dashboard. The cost tab has built-in intelligence to determine if your organization's cloud environment is over or under-provisioned from a sizing perspective. Although specific application scenarios can result in low utilization by design, you can often save money by managing the size and number of your virtual machine instances. Azure Advisor uses its artificial intelligence models to review your utilization history and can be made recommendations to lower your overall cost.
Along with virtual machine provisioning, storage provisioning is equally, if not more important than an application's performance. Storage can also have major implications on the cost of your Azure subscription.
First, ensure that all storage disks in your subscription are aligned to running active virtual machines. It is very easy to forget about a storage disk that was left behind after the removal of a virtual server. If you’re like many organizations with legacy-based data, start with a data-classification exercise to determine what data exists and where it currently resides within the Azure subscription.
Let’s start with a quick overview of Azure storage tiers:
Hot Access Tier
- Data that is in regular use or expected to be accessed (read from and written to) frequently.
Cool Access Tier
- Short-term backup and disaster recovery datasets.
- Older media content is not viewed frequently anymore but is expected to be available immediately when accessed.
Archive Access Tier
The archive access tier has the lowest storage cost, but it has higher data retrieval costs compared to the hot and cool tiers. Data in the archive tier can take several hours to retrieve and data must remain in the archive tier for at least 180 days or be subject to an early deletion charge. Long-term archival retention, along with data compliance that is required for a minimum of six months, is an excellent use case for the archive tier. Data stored in the cloud grows at an exponential pace. To manage costs for your expanding storage needs, it's helpful to organize your data based on attributes like frequency of access and planned retention period.
Azure Recovery (Backup) Vault
Review your backup retention within the Azure site recovery services vault to determine that the backup policy you are using aligns with the organization's RPO and RTO requirements. If they do not match, have a conversation with the business leaders or C-Suite to ensure they understand any potential gaps between the amount of backup retention you have and what the business requires.
Reserved & Burstable Instances
From Pay-As-You-Go to Azure Reserved Instances
It is in the best interest of an organization that has adopted the Azure platform to make further inroads to reduce overall operating expenses. Using Azure Reserved Instances, an organization makes a commitment on a 1-year or 3-year for a virtual machine's usage.
Unlike an up-front purchase where you pay the full amount, the monthly payment option divides the total cost of the reservation evenly over each month of the term. The total cost of up-front and monthly reservation costs is the same, and you do not pay any extra fees when you choose to pay monthly.
Reserved VM Instances are available for most VM sizes with some exceptions. Reservation discounts do not apply for the following VMs: A-series, Av2-series, or G-series. Under the Cloud Solution Program (CSP), of which iV4 is a Tier-1 provider, customers can elect to purchase reservations with monthly payments.
Use Burstable Instances
Converting servers that are not consistently busy, such as a domain controller or file server to burstable B-Series instances can allow an organization to save on the operational cost of the server. Web servers, small database servers, along with domain controllers often have a CPU characteristic that is very bursty from a performance standpoint. These workloads will run for extended periods using only a small fraction of the CPU performance potential. However, during "spike" usage periods, the CPU will then burst to meet application or track demands.
Without a burstable instance, while running in these low points, you are still paying for the full CPU, so that you can handle the high and bursty points. Effectively, you are spending money where you could be saving money. The B-Series offers a cost-effective way to deploy these workloads that do not need the full performance of the CPU continuously and burst in their performance. While B-Series VMs are running in the low points and not fully utilizing the baseline performance of the CPU, your VM instance builds up credits.
When the instance has accumulated enough credit, you can burst your usage up to 100% of the vCPU for the period when your application requires higher CPU performance. Burstable instances range in size from 1 vCPU and 1 GB of RAM to 20 vCPUs and 80 GB of RAM. If you're not using one of the six B-series class instances today, consider a comprehensive review of your monthly consumption.
If you’re already leveraging Azure, ProArch can help give your organization insight into your monthly consumption and determine how to potentially reduce costs. With a thorough analysis of cloud resources, we can help your organization meet cost objectives without sacrificing functionality.
As a result of the Azure cost optimization analysis, ProArch will provide:
- Detailed reports of Azure spending, broken down by service
- Cost optimization recommendations based on best practices and business objectives
- Security recommendations
- Identification of other remediation opportunities
Contact ProArch to manage your Azure spend and get the most from your cloud environment.