Tech ONTAP Blogs

FSx for ONTAP supplemental NFS datastores – Technical capabilities and enhancements journey


Last August, VMware announced NFS datastore support that allows independent scaling of compute and storage, enabling organizations to right-size their VMware Cloud on AWS SDDC deployment to control costs. NFS datastores supplement vSAN and can be used to store virtual machines, virtual disks, content libraries, ISO, etc. As this integration is built upon the long-standing NFS support used on premises, customers can leverage the enterprise features to simplify and optimize the VMware Cloud on AWS deployments. 


NFS datastore management is integrated into the VMware Cloud on AWS Console and API to simplify SDDC administration. Customers can simply add a datastore to one or more clusters. The service supports up to four datastores per cluster, and any datastore can be attached to every cluster in the SDDC. When using Amazon FSx for NetApp ONTAP as a datastore, Amazon Web Services (AWS) provides lifecycle management of the FSx for ONTAP file system (security updates, upgrades, and patches). VMware is responsible for the SDDC lifecycle management, and the customer is responsible for establishing the network connectivity, creating, and attaching the external NFS datastore to the SDDC. In this case, customer owns and manages the minimalistic storage configuration which provides more flexibility to the administrators. 


VMware Cloud on AWS supports both Single-AZ and Multi-AZ deployments of FSx for ONTAP. Before attaching FSx for ONTAP as a NFS datastore, first set up a VMware on Cloud SDDC environment or get an existing SDDC upgraded to v1.20 or above. For more information, see this document that walks through the steps necessary to provision and attach FSx for ONTAP as a NFS datastore for VMware Cloud on AWS. 


Connectivity Options

Since GA, Transit Gateway was required for Multi-AZ and Single-AZ file systems to access the datastore. While this integration helped the deployments, customers wanted to further reduce the total cost of ownership and infrastructure complexity. When introduced later this year, NFS datastores on a Single-AZ FSx for ONTAP file systems can be connected to SDDC clusters using VPC peering.  A VPC peering connection is a connection between two VPCs (in this case VMware VPC of the SDDC and customer VPC where FSx for ONTAP is running). To start with, the process is manual where in the customer will contact their Customer Success or Account Representative and request VPC peering. The owner of the requester (VMware) VPC sends a request to the owner of the accepter (customer) VPC to create the VPC peering connection. The owner of the accepter (customer) VPC needs to accept the VPC peering connection request to activate the VPC peering connection. The customer contacts VMware to finalize the VPC peering. VMware will then update the network routes to use the peering connection. The customer updates the routes in their VPC and configures any security group rules to allow NFS traffic. Once done, NFS storage traffic flows from the datastore mounted on the ESXi host through the VPC peering connection. This means Single-AZ file systems do not require Transit Gateway going forward. It simplifies deployment, improves latencies even further for FSx for ONTAP datastores that already had good performance, and reduces TCO when deploying VMware Cloud on AWS. The entire process will be automated in future releases further simplifying the process.  


With this enhancement, let’s look at the high-level deployment steps to mount a datastore to VMware Cloud on AWS SDDC:

  1. Create FSx for ONTAP in a new designated VPC.
  2. Configure VPC peering by accepting the request to activate VPC peering connection.
  3. Configure routing to use VPC peering and update the security groups to allow NFS traffic.
  4. Attach an NFS volume as a datastore to the SDDC cluster.

In case of Multi-Availability Zone FSx for ONTAP deployments, Transit Gateway is still required and instead of configuring VPC peering at Step 2, SDDC group along with Transit Gateway attachment should be used for attaching the datastores. However, majority of the customers prefer Single-AZ deployments as it meets both performance and cost for NFS datastores, unless multi-AZ level redundancy is required.




Increased throughput for NFS Datastores

Starting in the SDDC 1.22 release, the vSphere NFS client opens multiple network connections to each datastore mount automatically.  These connections are used on a round-robin basis and allow each vSphere host to increase the per datastore throughput. Earlier, each host maintained a single TCP/IP session for each datastore mount and now there are more TCP streams. This means that each datastore mount on each host can expect ~1GB/s throughput. Putting this together, each datastore can drive up to 1GB/s of throughput on each host. Hence, if a single datastore were mounted on four hosts, theoretically, the cluster would have access to 4GB/s of throughput.


A quick HCIbench test with a single host with four datastores mounted from a 4GB Single-AZ File system shows the effect of multiple sessions.




Cost Optimization

The NetApp TCO estimator, which uses VMC sizer as the foundation can be found here. An easy-to-use tool that offers insights into sizing aspects of the storage, simplifying and optimizing deployment. This estimator easily projects the no. of hosts that are required for a deployment and calculate the savings to optimize the deployment using Amazon FSx for NetApp ONTAP as a supplemental datastore to meet both performance and capacity requirements while complementing vSAN in a true hybrid storage model in the cloud.


Getting Started

  1. Open on a new browser.
  2. Select the Region and Sizer input method.
  3. Choose Manual input (equivalent to quick sizing). Another option is select Import RVTools or Live optics file for more accurate sizing based on collected data.
  4. Update the input parameters (Total number of VMs, vCPU/pCore, vCPU/VM, vRAM/VM, storage/VM.
  5. Update FSx for ONTAP parameters.
  6. Click Submit to generate the recommendation.


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





Demystifying Output table

The first table shows the direct comparison between ESXi hosts required vs FSx for ONTAP capacity to meet the requirements. So replace the ESXi hosts that is added for storage needs with FSx for ONTAP. In this demo example, instead of using 11 i4i hosts, 133TiB of FSx for ONTAP can be provisioned to achieve a TCO optimization of 92%. The second table shows the overall TCO comparison across vSAN only sizing vs reduced vSAN hosts + FSx for ONTAP to complement the deployment and optimize it.


Use the estimator to overcome the cost concern and challenge associated with VMware Cloud on AWS adoption.


Data Protection 

Apart from the TCO, FSx for ONTAP also exposes enterprise-grade features that customers have been using on premises within VMC SDDCs to optimise and streamline the deployment. In conjunction with vSphere, FSx for ONTAP Snapshot, SnapMirror, and SnapVault technologies enable the backup and restoration of virtual machines (VMs), datastores, and virtual machine disks (VMDKs). Backing up VMs and quickly recovering them are among the great strengths of FSx for ONTAP datastores. BlueXP backup and recovery for VMs is a Linux-based virtual appliance. Use Snapshot copies via BlueXP backup and recovery to make quick copies of the VMs or even the whole NFS datastore without affecting performance, and then send them to a secondary AWS region of choice using FSx for ONTAP SnapMirror replication for disaster recovery purposes.


What’s next - In a future release, like on-premises functionality, BlueXP backup and recovery for VMs will also work with SnapCenter Server to support application-based backup and restore operations in VMware Cloud on AWS SDDC environments for SnapCenter application-specific plug-ins.


For detailed information covering the deployment method of protecting FSx for ONTAP NFS datastores and the associated VMs using NetApp BlueXP backup and recovery for VMs, please refer to the blog here.


Disaster Recovery

Another key capability is NetApp SnapMirror. Disaster recovery using block-level replication from on-premises to VMC SDDC or between SDDC regions within the cloud is a resilient and cost-effective way of protecting the workloads against site outages and data corruption events, like ransomware attacks. With SnapMirror replication, VMware workloads running on-premises ONTAP systems using NFS datastore can be replicated to FSx for ONTAP in a designated AWS target recovery region where VMware Cloud on AWS SDDC resides.


Based on this recent AWS blog post that covered the benefits of using SnapMirror for disaster recovery, a community-supported scripted solution Disaster Recovery Orchestrator (DRO) was released to overcome the siloed scripts approach. Customers are using DRO to meet DR scenarios. If you are a scripting geek, DRO is there for you. For more information, see Disaster Recovery with FSx for ONTAP and VMware Cloud on AWS.


Now based on huge customer demand, we are also happy to introduce the same functionality called NetApp BlueXP disaster recovery that is added to the BlueXP control plane to simplify business continuity. Using the BlueXP disaster recovery service, which is integrated into the NetApp BlueXP console, customers can discover their on-premises VMware vCenter and AWS SDDC vCenter along with FSx for ONTAP, create resource groupings, create a disaster recovery plan, associate it with resource groups, and test or execute failover and failback.


For more information, see BlueXP disaster recovery service for business continuity with VMware Cloud on AWS and Amazon FSx for NetApp ONTAP.



One of the other most common use cases is migration. Customers can use VMware HCX, BlueXP disaster recovery migration option using SnapMirror or other third-party migration tools to migrate VMs.


For additional information about migration options and on how to migrate workloads from on-premises to VMware Cloud on AWS, see VMware Cloud TechZone, VMware HCX User Guide, and Migrate workloads to FSx for ONTAP datastore using VMware HCX.



This blog covers the enhancements added over the last year to Amazon FSx for NetApp ONTAP as a datastore with VMware Cloud on AWS. FSx for ONTAP provides excellent options to deploy and manage application workloads along with file services while reducing the TCO by making data requirements seamless to the application layer. Whatever the use case or application, choose VMware Cloud on AWS along with Amazon FSx for NetApp ONTAP for rapid realization of cloud benefits, consistent infrastructure, and operations from on premises to AWS, bidirectional portability of workloads, and enterprise-grade capacity and performance. It is the same familiar process and procedures used to connect storage. Remember, it is just the position of the data that changed along with new names; the tools and processes all remain the same, and FSx for ONTAP helps to optimize the overall deployment.