Cloud Volumes ONTAP (CVO) is a software defined storage (SDS) platform from Network Appliance that runs in the public cloud. It’s supported in all three major public clouds, AWS, Azure, and GCP and is one of two cloud offerings from NetApp. The other is Cloud Volumes Service which is a fully managed file service, or PaaS for file.
In this series we’ll focus on Cloud Volumes ONTAP in AWS. As with most IaaS, when compared to PaaS, it has more administrative overhead, but provides more features and flexibility. One of those features, replication, is quickly becoming the most common use case for its adoption.
NetApp’s ONTAP operating system revolutionized enterprise NAS storage over 20 years ago. It went on to support block protocols and became a unified platform – a Swiss army knife for storage. Their appliances are now serving data in tens of thousands of customer data centers worldwide.
Cloud Volumes ONTAP is the same platform that runs on physical hardware in the data center, but it runs in the public cloud as an instance or VM. This is not a new concept exclusive to NetApp. Most storage vendors that have relied on traditional on-prem, hardware based deployment models have been forced to adjust to the reality of increased cloud adoption.
Several prominent storage vendors have released and touted the promise of their cloud based solutions recently. Some have taken a relatively simple approach to cloud integrated features by enabling tiering to S3 compliant cloud storage from their physical arrays. Netapp’s FabricPools and Isilon’s CloudPools are examples of this.
While these features are useful, they’re often mistaken for data protection solutions. They allow customers to extend the capacity of their on-prem storage appliances by tiering to the cloud, but they do not create a second independent copy of the data. If your on-prem appliance is lost due to a local or regional disaster, so is the data. The S3 bucket which remains in tact after the disaster is useless because the metadata required to access it was lost along with your appliance.
Qumulo and Cohesity, two enterprise grade NAS platforms that have software-defined storage options in the cloud, are worth acknowledgement here. Both have their strengths in cloud integration. Qumulo scales to support very large data sets with ease and Cohesity offers advanced data reduction and data protection for a variety of workloads. Notably absent in this space is Dell EMC’s Isilon. While the Isilon on-prem scale-out platform is one of the most capable in the market and can tier to the cloud, it can’t actually run as an SDS instance in the cloud today.
Common use cases for storage platforms in the cloud are bursting, data reduction, and data protection.
Bursting is enabled by using native, block-based replication between an on-prem and cloud based storage platform to bring data closer to cloud compute. This typically makes sense when the output can be consumed or analyzed in the cloud directly or is small enough to avoid significant egress times and cost.
Data reduction is really a cost optimization method. If a cloud storage platform can deduplicate and compress a data set it will reduce cloud storage consumption and ultimately cost. The catch is that the savings must exceed the licensing cost of the cloud storage platform by enough to justify the added complexity of inserting it in between your apps and cloud storage.
Data protection is probably the most straightforward use case. Protocol matters. Although it’s possible to present block storage to other instances in the cloud it can be a challenge, especially when the source involves a hypervisor like VMware’s vSphere. Software iSCSI initiators and VMware’s vVols make it possible, but they introduce layers of complexity many DR plans aim to avoid.
NAS data protection on the other hand can be simple and reliable. It can enable recovery of data sets in the cloud to serve apps running in the cloud, in data centers, or directly to remote users.
In this series we’ll examine Cloud Volumes ONTAP in AWS for this use case. We’ll focus specifically on cost optimization and workload design. As with any cloud based solution, while it might be easy to deploy and maintain, it must be cost effective for it to make sense as a part of an enterprise’s data protection scheme. In the next post, we’ll focus on optimized selection of AWS storage for use with CVO.