Software licensing: The blind spot in public cloud costs

Over the last five years, as enterprise IT teams have implemented better cloud governance, they have moved from “cloud bill shock” to actively managing and optimizing their public cloud spend. In fact, the RightScale 2018 State of the Cloud Report found that optimizing cloud costs is the top initiative for the second year in a row for all cloud users.

However, the additional costs of licensed software running on public cloud instances is still a blind spot for many organizations. In some cases, the cost of the licensed software may far exceed the cost of the cloud infrastructure on which it runs. Cloud has also become a hub for applications that connect to edge devices for mobile and IoT. These edge devices may also run licensed software components that need to be included in the overall cost of a service. Without a complete picture of all of the costs of cloud workloads, including software license costs, organizations risk ballooning costs caused by un-optimized spend.

While enterprises are leveraging PaaS services (such as database-as-a-service or Hadoop-as-a-service) for some workloads, they are also frequently installing licensed software on cloud instances. Those licenses may be covered by existing enterprise agreements or they may represent net new licenses that need to be purchased.

Because leading software vendors often price software licenses differently for public cloud instances, the software running in the public cloud (even in the same-size virtual machine or on the same CPU chipset) may not cost the same as when running on-premises. In addition, there are often incentives provided by software vendors to run the software on the vendor’s own public cloud. All of these factors should be considered whether you are migrating existing workloads to the cloud, building greenfield cloud applications, or negotiating pricing.

Let’s look at two examples that illustrate the importance of considering software license costs in the cloud.

Example 1: Oracle Database Enterprise Edition licensing

Oracle maps CPU cores to licensed processors differently between on-premises and cloud computing environments. As a result, the software license costs for Oracle Database Enterprise Edition will be twice as much in AWS or Azure as for on-premises deployments. Oracle is clearly pushing customers toward its own cloud or on-prem.

There are several options you can consider to mitigate the inflated Oracle Database pricing on AWS and Azure clouds.

• Rightsize your instance. Make sure you are not over-provisioning vCPUs in the cloud. If your limiting factor for performance is memory, you can pick instances that offer the highest ratio of memory to vCPU, such as AWS X1 instances or Azure G-series or M-series instances.
• Limit the vCPUs allocated. Even after picking the right instance sizes, you may want to further limit the number of vCPUs to control the cost. The AWS Optimize CPU and Azure Constrained CPU options let you reduce the vCPUs and hence reduce the software license cost.
• Consider database-as-a-service offerings. You may want to consider whether other database-as-a-service offerings would meet your needs. AWS and Azure both offer multiple SQL database engines including MySQL, MariaDB, PostgreSQL, and their own proprietary offerings.

Example 2: Azure Hybrid Benefit (AHB)

Microsoft allows you to apply unused licenses for Windows Server or SQL Server to your Azure instances, and thereby leverage software licenses you have already purchased to reduce the hourly cost of your Azure instances.

All Azure users should plan to take advantage of the Azure Hybrid Benefit to optimize their Microsoft spend.

• Review your Microsoft license position. For Windows Server Datacenter Edition, you can use each license twice—simultaneously on-premises and in Azure. For Windows Server Standard Edition, you can take unused Windows Server licenses and re-allocate them to be used in Azure.
• Pay attention to vCPUs. Each two-processor license for Windows Server is entitled to two Azure instances of up to eight cores, or one instance of up to 16 cores. You should apply AHB to instances of eight or 16 cores whenever possible to maximize the benefit.
• Implement continuous license optimization. Applying the AHB is not a “once and done” activity. You will want tools to continually monitor the licenses being used both on-premises and in the cloud to ensure you are not overbuying. As on-premises license use rises, you will need to release licenses assigned to AHB in Azure and allow those Azure VMs to revert to on-demand pricing for Windows. This approach helps you optimize the use of your licenses while staying in compliance.

Software costs matter in the cloud

These are just two examples that illustrate the cost and licensing impacts of software assets in the public cloud. As you move to the cloud, you have the opportunity to take new approaches to managing software licenses — by rightsizing your cloud resources, re-allocating licenses between on-premises and cloud environments, and continually optimizing the use of software licenses you’ve already purchased.

Kim Weins is vice president of cloud strategy at Flexera. She has held executive marketing positions at a variety of enterprise software startups and public software companies. She received a B.S. in engineering from Duke University.

—

New Tech Forum provides a venue to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to newtechforum@infoworld.com.