TCO Estimator - Guide to optimizing VMware Cloud on AWS deployment using FSx for ONTAP Datastores

1. Open https://bluexp.netapp.com/vmc-aws-fsx-ontap/roi on a new browser.
2. Select the Region and Sizer input method.
3. Choose Use RVTools File and click on Choose file to upload RVtools file.

   Note: If RVtools cannot be uploaded due to data sharing policies, use the parser available which parses the data within your network and the same can be used for sizing by selecting the “Manual Sizing” option.

4. Update the considerations for parsing RVtools.
   1. The recommendation is to select Powered ON and OFF VMs.
   2. The storage to consider should be Utilized. Utilized means total capacity used measured across all drives on the guest VMs. Provisioned can be selected depending on the requirements.
   3. The memory to consider for sizing should be Utilized
5. Update the FSx for ONTAP storage parameters.
   1. Deployment type – Single AZ by default

      Note: Majority of the deployments use Single AZ FSx for ONTAP file systems.
      Note: FSx for ONTAP is not supported with stretched clusters

   2. Define the workload profile. The approach here is to size the file system for the appropriate IOPs and throughput based on app specific workloads. Since any kind of data collection tool doesn’t provide default workload annotations, a guesstimate is used here. The assumption is 20% of data stored on FSx for ONTAP datastore will be database workload and the remaining 80% is generic workload. This can be modified to suit specific customer environment needs. If no additional data point exists, use the default values.

      Note: This part requires more attention. It provides two different profiles (database and virtual machine data as default profiles), but these could be any majority workloads in the customer environment, such as VDI and generic VM workloads, or even 100% general VM workloads

   3. Percentage of data on SSD storage

      The recommendation is to keep 100% on SSD. However, if the workload data involves cold data (purely based on data points and inputs from specific app owners). Tiering should be enabled only for snapshots for Tier 0 and Tier 1 NFS workload datastores. For the rest, tiering can be set to Auto.

   4. Savings from compression + deduplication

      The default value is 65% for generic VM workload and can range upto 80%. This value is derived based on the efficiency gains seen by on-premises customers. While generic VM workload can see significant benefits, database workloads would provide lower savings, hence consider any value between 15-25%.

      If sizing needs to be conservative, then consider 40%.

   5. Provisioned SSD IOPS

      Modify the SSD IOPs based on the IOPs required. The IOPs can be modified anytime to the max achievable for each SKU type depending on the requirements. Amazon FSx for ONTAP automatically provisions 3 SSD IOPs per GB for the file system, up to a maximum of 160,000 SSD IOPS per file system. Keep in mind, the maximum level of SSD IOPS that a file system can achieve is also dictated by the file system's throughput capacity, even when provisioning additional SSD IOPS.

   6. Throughput capacity

      Modify the throughput capacity to suit the needs. If no data is available, use 1GB SKU. Remember, the throughput capacity can be modified on the fly with a single click. If the workload access pattern is known and performs random reads, then 2GB or 4GB can be specified to take advantage of flash cache capabilities for Single AZ file systems. If using Multi-AZ, flash cache is available for all SKUs.

6. Disaster Recovery Profile
   1. Percentage of data on SSD storage

      The recommendation is to keep 50% on SSD and use the Auto tiering policy.

   2. Savings from compression + deduplication

      The default value is 65% for generic VM workload and can range up to 80%.

   3. Provisioned SSD IOPS

      Modify the SSD IOPs based on the IOPs required. The IOPs can be modified anytime to the max achievable for each SKU type depending on the requirements once the DR is activated.

   4. Throughput capacity

      Modify the throughput capacity to suit the needs. For DR usecases, keep the throughput to a smaller SKU and then upgrade when disaster event is triggered.

   
7. Transit Gateway Attachment

   For Single AZ deployments, VPC peering is used and hence TGW calculations are disabled. In the case of Multi AZ deployments, data flows from the datastore mounted on the ESXi host within VMC SDDC via the VMware Transit Connect. So TGW calculations will be enabled. If “Auto calculate” is used, the estimator considers 20% of the total FSx for ONTAP capacity as the change rate and numbers are auto calculated based on it.

8. Select the savings options that should be used in the calculations.
9. Click Submit

Once the sizing request is submitted, the payload is sent to the VMC sizer and TCO calculations are applied on the projected numbers.

For manual sizing , refer to the documentation.