Migrating databases such as Microsoft SQL Server from an on-premises deployment to Amazon Web Service (AWS) can be a major undertaking. There are many moving parts to consider, one of the most important being how to balance compute costs and performance.
There are many Amazon Elastic Compute Cloud (Amazon EC2) types available, presenting a range of performance and cost options for your database requirements. But it's important to consider the entire infrastructure stack for your database deployment on AWS. The choice of one component can have an impact on another. For example, your choice of storage can impact the Amazon EC2 instance type and sizing, forcing you to select a more performant—and expensive—instance than your database requires.
In this post, I’ll show how Amazon FSx for NetApp ONTAP (FSx for ONTAP) can help you optimize your Amazon EC2 costs while delivering the performance levels your database needs. We'll explore the challenges of database migration, share a customer success story, delve into the technical aspects of FSx for ONTAP, and provide a detailed total cost of ownership (TCO) calculation.
Here’s what we’ll cover:
- The compute selection challenge
- Customer story: A SaaS company’s SQL Server migration
- How FSx for ONTAP helps optimize your Amazon EC2 type selection
- Illustrating the cost benefits of FSx for ONTAP
- Amazon EC2 + Amazon EBS configuration
- Amazon EC2 + FSx for ONTAP configuration
- The TCO analysis conclusion
- Simplify migration planning with NetApp BlueXP workload factory for AWS
- Conclusion
The compute selection challenge
Migrating databases to AWS provides a long list of benefits, but there are some points that need to be addressed, such as availability and resiliency, and managing the migration process itself. One of the most critical aspects is selecting the Amazon EC2 instance type from the wide range of choices that AWS provides. You need to find the type that will provide sufficient performance for your database while remaining cost effective at scale.
There are two parts to selecting your compute resource on AWS:
- Amazon EC2 instance type selection.
There are many Amazon EC2 instance types available: general purpose, storage optimized, compute optimized, memory optimized, accelerated computing instances, etc. Each of these caters to specific workloads and use cases.
- Sizing the instance.
After you choose an instance type, you need to select the size of the instance. There is a range of sizing options specific for each instance type. For example, storage optimized instance sizing options include i8g.2xlarge, i7i.large, i16g.2xlarge, etc. Generally speaking, the larger the instance, the higher the performance level and cost.
For more details, see this chart of Amazon EC2 instance types and sizes.
Selecting the instance type and size that best fits your database in terms of cost and performance is already a challenge. But there is another factor to consider: Storage type and sizing. When choosing Amazon EBS volume types, specific Amazon EC2 instance types may be required. That can potentially lead to configurations that aren’t cost optimized and provide higher performance than your workload requires.
Customer story: A SaaS company’s SQL Server migration
Here's a real-world example of a software-as-a-service (SaaS) company that successfully migrated its SQL Server workloads from its data center to AWS. The company had 20 databases requiring 32 vCPU cores, 256 GiB of memory, 100,000 input/output operations per second (IOPS), and 5 TB of data capacity. As they considered deployment options on AWS, they compared two services: Amazon EBS and FSx for ONTAP.
To identify the most suitable Amazon EC2 instance type, the company focused on memory-optimized types, because SQL Server databases generally benefit from large amounts of RAM while requiring relatively modest compute resources. After careful consideration, r6i.8xlarge seemed to meet the company’s compute requirements, offering 32 vCPU cores and 256 GiB of memory.
However, a challenge emerged regarding the 100,000 IOPS requirement. Upon reviewing Amazon EBS-optimized instance types, they discovered that r6i.8xlarge could deliver a max of 40,000 IOPS. To meet the 100,000 IOPS requirement, the company would need to upgrade to r6i.24xlarge.
This posed a significant issue: r6i.24xlarge was much more expensive and much more performant than the company’s database needed. Because of their IOPS requirements, the company would be forced to select an instance type that would exceed their workload requirements. r6i.24xlarge provides 96 vCPU cores and 768 GiB of memory—all of which comes along with a much higher price tag.
The company then evaluated using FSx for ONTAP.
They found FSx for ONTAP could provide a cost-effective storage solution that would optimize performance and allow for flexible Amazon EC2 instance selection, thereby reducing overall costs. By leveraging network bandwidth, FSx for ONTAP allows the choice of Amazon EC2 instance type to be decoupled from the storage layer. The company was thus able to use r6i.8xlarge while achieving the optimal cost-performance metrics from its storage volumes.
According to the FSx for ONTAP guide's performance section, the company could attain 100,000 IOPS and 3,072 MBps throughput—decoupled from the Amazon EC2 instance type.
FSx for ONTAP contributed to additional savings in two other ways. It reduced the total number of Amazon EC2 cores required—thus lowering the SQL licensing costs—and it shrank the total data footprint through its unique storage efficiency and thin cloning technologies. Read more about lowering your database costs with FSx for ONTAP.
Ultimately, the company achieved a 40% reduction in compute costs and a 65% decrease in the total cost of ownership for their workload, all without compromising performance requirements.
How FSx for ONTAP helps optimize your Amazon EC2 type selection
Amazon FSx for NetApp ONTAP is a fully managed AWS storage service with the benefits of NetApp® ONTAP® baked in. It has high-performance characteristics that are essential for demanding workloads, such as databases:
- High throughput file systems
- SSD disks with sub-millisecond latencies
- Up to 200,000 IOPS
- In-memory cache and nonvolatile memory express (NVMe) caching to further enhance the performance of frequently accessed data
FSx for ONTAP also uses network bandwidth—one of the two interfaces Amazon EC2 uses to connect to the underlying storage, the other being Amazon EBS bandwidth. Using network bandwidth allows the performance characteristics of the FSx for ONTAP system to remain independent of the associated Amazon EC2 instance type in use. This separates the choice of storage from the choice of instance type, meaning you can optimize the resources you utilize to get the best cost-performance.
Illustrating the cost benefits of FSx for ONTAP
To highlight the cost advantages of FSx for ONTAP, we conducted a detailed TCO analysis for the SaaS company in our story. This is a helpful example to demonstrate the value of FSx for ONTAP in general, so I’ll present our findings here in full.
Note: The performance characteristics used in this calculation were current as of Q2 2025.
Amazon EC2 + Amazon EBS configuration
First, we calculated the costs for the company’s initial configuration using Amazon EC2 and Amazon EBS. The following parameters were used in the TCO calculation:
- Tenancy: Shared instances
- Operating system: Windows Server
- Amazon EC2 instance type: r6i.24xlarge (96 vCPUs, 768 GiB memory)
- Number of instances: 2 (for resiliency)
- Pricing strategy: 3-year, No upfront
- Amazon EBS storage amount (per io2 volume): 5 TB
- IOPS: 100,000
The estimated cost for 12 months was $260,384.16.
Amazon EC2 + FSx for ONTAP configuration
Next, we calculated the revised, FSx for ONTAP-based deployment. The Amazon EC2 cost in this setup involves only a small Amazon EBS volume to host the operating system:
- Tenancy: Shared Instances
- Operating system: Windows Server
- Amazon EC2 instance type: r6i.8xlarge (32 vCPUs, 256 GiB memory)
- Number of instances: 2 (for resiliency)
- Pricing strategy: 3-year, No upfront
- Amazon EBS storage amount (gp3): 50 GB
- IOPS: 120,000
The cost for Amazon EC2 and minimal Amazon EBS storage is $44,310 for 12 months.
Adding FSx for ONTAP to this configuration:
- Deployment type: Multi-AZ
- Disk type: 100% SSD
- Savings from compression and deduplication: 0%
- Aggregate throughput: 3,072 MBps
- Storage capacity: 5 TB
- Aggregate SSD IOPS: 100,000
The cost for FSx for ONTAP is $94,129.92 for 12 months.
The combined total for this deployment is $138,439.92.
The TCO analysis conclusion
This analysis demonstrates a 47% cost savings when using FSx for ONTAP. These savings stem from optimizing the choice of Amazon EC2 instance size, enabling the use of smaller Amazon EC2 instance types with reduced memory and vCPU requirements.
Additional savings are realized through lower database licensing costs for smaller Amazon EC2 instance types, while FSx for ONTAP further reduces storage expenses with built-in storage efficiency technologies—in particular, thin cloning, which makes it possible to create instant copies of the database without noticeably increasing costs—indicating that the savings actually exceed 47%.
Simplify migration planning with NetApp BlueXP workload factory for AWS
Now that we understand how FSx for ONTAP can optimize your compute requirements, there is another tool that can simplify your database migration planning: BlueXP™ workload factory for AWS. Workload factory is a free service that automates the FSx for ONTAP deployment process, so your instances and storage volumes are aligned with workload-specific requirements and best practices right out of the box. Workload factory offers an on-premises SQL Server migration advisor to help you see in advance what migrating your database to AWS will look like, including the way it will work with FSx for ONTAP storage once it gets there.
The workload factory SQL Server migration advisor helps to plan the migration and translate the on-premises configuration to an optimal AWS deployment, including the Amazon EC2 instance selection. With this information—along with support for different source environment systems, an overview of your projected TCO, and optimization recommendations for your workload—your team can make the best choices for your workload to thrive on AWS.
Here’s a look at the flow of the workload factory migration advisor for on-premises databases:
Learn more about BlueXP workload factory or get started now.
Conclusion
Migrating databases to AWS is a major move for any organization. FSx for ONTAP and BlueXP workload factory make it easier. helping you optimize both costs and performance.
By separating storage from compute concerns, FSx for ONTAP allows you to use smaller Amazon EC2 instances, leading to significant cost savings. With BlueXP workload factory, you can simplify the migration planning process, getting comprehensive assessments and optimized configuration recommendations to make your database’s migration to AWS a success from the start.
Learn more about FSx for ONTAP here or request a meeting with a specialist.
