In the previous post The 5-R Modernization Strategies: The Road to REHOST, we explored the motivations that led to a public school district choosing to rehost server workloads in Azure. In this post, we’ll provide an overview of how we used Azure Migrate to prepare for the migration.
Before migration efforts began, we needed to thoroughly assess the workloads being migrated. This assessment helped us determine important information like the Azure hosting costs, what workloads and user communities were going to be impacted, and how the workloads needed to be migrated.
First, we leveraged Azure Migrate to flag server-level problems that would require more consideration before a migration could occur, such as servers using unsupported Operating Systems. Those flagged servers would require us to perform additional tasks such as upgrading the Operating System or performing the migration in a different manner.
We also used Azure Migrate to calculate utilization metrics for RAM, CPU, and storage to a standard deviation. This utilization assessment ran for two weeks to ensure we received the most accurate results. This assessment also provided recommendations to us for the right Azure VM size for a given server based on the utilization marks
In KiZAN’s experience with these assessments, customers significantly underutilize their on-premises hardware to the tune of 70% (in many cases, the underutilization numbers are even more egregious).
A large underutilization number makes sense when considering traditional self-hosting models must account for peak utilization of their workloads and not just the average day-to-day usage.
With the economies of scale that Azure’s datacenters provide, this consideration changes and allows for us to “right-size” the virtual machines closer to their actual utilization numbers. In the case of the public school district, this right-sizing effort provided some cost savings.
There was one other area where Azure Migrate provided KiZAN critical information.
In modern IT environments, workloads are no longer standalone applications on a couple of servers – rather we tend to find a highly entangled environment where multiple applications (and dozens of servers) are in communication with each other. This complicates the migration path. Azure Migrate has a feature that can be configured known as dependency mapping.
There are two different methods for configuring the feature, agent-based or agentless.
Agentless only works with VMWare hypervisors while agent-based works with multiple hypervisor technologies. This feature works by capturing TCP connection data in five-minute intervals. The resulting data gave us source and destination server information and helped us properly group servers that were in communication with each other. We could then assess these servers in groups and categorize them into migration waves.
Save money and bring your workloads to the cloud with our certified best practices.
KiZAN is a Microsoft National Solutions Provider with numerous gold and silver Microsoft competencies, including gold data analytics. Our primary offices are located in Louisville, KY, and Cincinnati, OH, with additional sales offices located in Tennessee, Indiana, Michigan, Pennsylvania, Florida, North Carolina, South Carolina, Georgia and Texas.