February 24, 2020
Topics: Cloud Tiering Data TieringAdvanced6 minute read
NetApp has a variety of AFF models to suit the IT demands of companies that range from medium-sized business to large enterprises. With the recent addition of the C190, there is now even a model for startups and other small businesses.
And while these are all dedicated on-prem storage systems, that doesn’t mean they can’t take advantage of the economics and scale of the cloud. To save on data space, and future outlays for new storage systems, all of NetApp’s AFF offerings can leverage the Cloud Tiering service to store infrequently accessed data in AWS, Azure, or Google Cloud.
In this article we will cover each of NetApp’s AFF models, explain more about the technology, and finish by showing you the benefits of using the Cloud Tiering service with them.
What Is AFF Storage?
AFF stands for All-Flash FAS. FAS is Fabric Attached Storage, which means that block and file storage can be presented over multiple types of networks. This makes AFF a purely flash storage system for NAS and SAN applications.
All AFF systems run the NetApp ONTAP data management software which gives feature parity across the whole range, so you do not have to compromise, you only need to decide on performance and capacity.
With ONTAP comes System Manager. System Manager is a graphical management interface that provides a consistent look and feel when managing any ONTAP cluster and is used to manage all your configurations, whether it is an additional network, sharing via NFS and SMB, or iSCSI and Fibre Channel. Even backup and other resilience features such as SnapMirror® data replication are all managed through the interface, with wizards to optimize time spent provisioning storage and other tasks.
A single AFF system chassis contains two AFF nodes configured as an active/active HA pair with both having access to all discs, therefore if one node does fail the other node will take over its work. Furthermore, a single HA pair can form an ONTAP cluster or be a part of a larger array that can scale up to 24 AFF nodes.
Connectivity varies across the range, but each system will have multiple high bandwidth network interfaces for serving data over Ethernet, Fibre Channel and Converged Ethernet networks such as: 100Gb, 40Gb, 25Gb, or 10Gb Ethernet, or FC 32Gb or 16Gb.
NetApp’s AFF Offerings
In this section we’ll give a capsule review of each of the AFF options NetApp has to offer.
This is the speedster, the fastest performer of the range designed for workloads requiring ultra-low latency. If you are doing data analytics, AI, deep learning or any other workload that requires a consistent large amount of IOPS with sub-millisecond response time, storage performance will not be your bottleneck with this AFF. Also, with a maximum physical capacity of ~6PB per HA pair and an average of 5:1 storage efficiency ratios, you will not have to compromise on capacity.
For ultimate performance, the A800 uses NVMe SSDs and supports NVMe/FC, that impressively increase the amount of IOPS, decrease latency and host CPU utilization. This is the model for the use cases that demand the highest performance, and as such it is also the most expensive option on this list.
This was previously the highest performing enterprise storage system before the A800 premiered, and it is still a very performant system. The CPU and memory specifications are very similar, but the A700 has twice the storage capacity of the A800 and was designed for cases where a modular architecture is required to meet future workloads and data center demands.
Within the 8U highly modular chassis up to 64 x 32Gb FC or 32 x 40Gb ethernet interfaces can be installed for great connectivity and from capacity perspective, this AFF model can scale up to ~14PB per HA pair.
The balance of price-performance and great capacity makes the A400 a great mid-range storage system suitable for most enterprise applications. Similar to the A800 model, the A400 can be expanded using NS224 NVMe disc shelves and the support for NVMe/FC and NVMe/RoCE. That enables A400s to achieve the performance, scale and operational efficiency goals of AI/ML, real time analytics, and NoSQL workloads.
The A220 is at the lower end of the scale and it addresses the mid-size businesses and small enterprises requirements for simplicity and value. At 2U it can hold 24 SSDs internally and can be expanded with external disc shelves to a maximum of 144 SSDs per HA pair, which equates to a maximum raw capacity of ~4PB per HA with ~16PB of effective capacity based on the use cases used.
The new C190 AFF unit is ideal for small businesses or remote offices of an enterprise. The C190 runs the same ONTAP software but has 2 limitations, it does not support metro-cluster, and cannot be expanded by additional shelves thus is limited to 24 x 960GB SSDs.
The C190 can deliver data simply, quickly and cost effectively while allowing IT modernization with an affordable all-flash storage solution that has a maximum effective capacity of ~50TB based on a 3:1 efficiency ratio (actual ratio can be higher depending on the workloads and use cases).
|Best fit||Small Office or satellite office||Mid-size business and small enterprises that require simplicity and best value||
Most enterprise applications that require best balance of performance and cost
Performance-driven workloads and data center requiring modular design
Most demanding workloads requiring ultra-low latency
|Max cluster size ||24 nodes||24 nodes||24 nodes||24 nodes||24 nodes|
|Max SSD Per HA Pair||24||144||480||480||240|
|Max SSD per Cluster||288||1728||5760||5760||2880|
|Max Effective Capacity||50TB||193.3PB||702.7PB||702.7PB||316.3PB|
The Added Benefit of Cloud Tiering
As different as the models of the range are, as AFF systems running ONTAP, they can all use the new Cloud Tiering service.
Cloud Tiering makes it possible to classify data blocks as hot or cold, based on their usage, and those classified as cold are automatically moved out to object storage in the cloud in Google Cloud, AWS, or Azure. Once that data is required again, Cloud Tiering moves the blocks back to the AFF system automatically and seamlessly.
This reduces the storage capacity requirements on-prem and can reduce the requirement for expanding your existing storage or reduce the required capacity of new storage system purchases. This reduced capacity can result in large cost savings; with some storage costs able to be moved into the operational expenditure budget, a further reduction in capital expenditure can be expected. Read about all the use cases for this technology here.
Cloud Tiering has three storage tiering policies that target all inactive data, snapshot data, or entire volumes that can be better stored in the cloud at low cost than on the AFF system. The reclaimed space can be reserved for applications and the most latency-sensitive or critical Tier 1 data. For verticals as different as media & entertainment, education, and retail, this can be an extremely useful way to reduce costs, save space, and more.
For more on Cloud Tiering and how to start using it with your AFF system, click here.
 24 Nodes made up of 12 HA Pairs
 Effective capacity based on 5:1 storage efficiency ratios with the maximum number of SSDs installed; space savings will vary dependent on workload and use cases.
 Effective capacity based on 3:1 storage efficiency ratios with the maximum number of SSDs installed; space savings will vary dependent on workload and use cases