File-based cloud storage is now a standard offering from Amazon Web Services, Google Cloud Platform and Microsoft Azure. Although the cloud giants’ offerings are most often associated with object storage, all three, along with suppliers such as IBM and NetApp, have developed offerings that allow businesses to use familiar hierarchical file storage in the cloud.
File-level storage in the cloud is either a native offering built into the hyperscaler’s technology stack, a partnership with a specialist supplier such as Azure NetApp Files, or what analyst IDC describe as “overlay” file services. These use software to present the hyperscaler’s block or object storage as a file system, usually SMB or NFS. Examples include Elastifile, now part of Google, and NetApp Cloud Volumes ONTAP.
Each approach brings its own benefits, and allows IT teams to fine-tune cloud storage according to their use cases, business requirements, and budgets. But, although all three cloud giants now support file-level storage, there are differences that make some options more suitable for some use cases than others.
The big three cloud providers’ core storage infrastructure is based around object storage technology. Meanwhile, however, the majority of on-premise storage is either block or file, either direct-attached to servers or in SAN or NAS shared storage (although more suppliers offer on-premise object storage nowadays).
The hyperscalers favour object storage because it meets their business needs. For now, customers that need the specific features of object storage are most likely to use it in the cloud. Only a minority can justify on-premise object storage.
Instead, most businesses continue to run file storage on-premise, often in NAS environments. And so, operating a consistent, file-based environment locally and in the cloud will cut management overheads, and reduce the need to build new skillsets.
Ease of management should not be underestimated. Integrating cloud-based storage adds complexity to IT architecture so being able to use familiar, file-based storage helps. A consistent file structure also makes it easier to move workloads to or from the cloud, even if the organisation stops short of full hybrid working.
Cloud file storage pros: Application compatibility
Most enterprise applications are not – yet – fully compatible with object storage. Support for S3 and the RESTful API is growing, but file-based storage will work out of the box.
With file storage, organisations should be able to migrate workloads to the cloud and back again without the need to rewrite software. This lift-and-shift approach is only really possible if the underlying file structure is the same.
Potentially, it makes disaster recovery easier, too. An application can be stood up using cloud-based backup data without the need to use object-storage protocols or go through an overlay service, with its potential compatibility and performance issues.
Using file storage also makes it easier to share data between applications or custom processes that expect a hierarchical file structure.
Cloud file storage pros: Support for hybrid environments
Object storage will work in hybrid cloud environments, and suppliers are investing in on-premise object technologies that make it easier. Software-defined storage is also often object-based.
Even so, using this technology requires expertise and investment. Businesses are most likely to use hybrid cloud object storage environments if they need to hold or process extremely large datasets or need to replicate data across geographies.
A hybrid environment based on a file structure, usually NFS or SMB, allows the business to extend its NAS capacity into the cloud, and to buy capacity on demand. Local volumes deliver performance, while the cloud adds additional, low-cost capacity.
NAS technology suppliers increasingly support tiering to the cloud, as well as use cases that include backup and archiving. The advantage of a file-based hybrid environment, however, is that no special technology is needed on the end user side. It should just work, with apps – and users – able to access data on-premise and remotely.
AWS, Azure and Google all offer native cloud NAS support, as well as volumes based on NetApp technology.
Cloud file storage pros: Support for low-latency use cases
File-based storage is better than object storage for applications that need low latency and high data throughput. As IDC points out, file storage services are tightly integrated with providers’ other resources, such as compute so can help maximise performance.
This contrasts with object-based storage, which uses its ability to scale to deliver services such as archiving at lower costs.
Cloud file storage pros: Availability of different service levels
Organisations can use file storage in the cloud to take advantage of more flexible cost structures than on-premise storage can offer.
An organisation that invests in an enterprise NAS usually needs to specify hardware based on the performance needs of the most demanding applications.
In the cloud, however, buyers can pick performance levels, SLAs and even file systems according to their needs. AWS, for example, offers Standard and Infrequent Access tiers for its EFS service.
The hyperscalers’ cloud file storage options
AWS’s file-based offering is Elastic File System (EFS), an NFS-based architecture that runs in the cloud and on local storage. The supplier also offers FSx for Windows File Server for users that prefer a dedicated Windows option. Users can automate tiering between Standard and Infrequent Access volumes.
Azure Files runs the SMB protocol, again both in the cloud and locally. It supports Windows, Linux and more unusually, MacOS. Volumes can be up to 4PB. Unsurprisingly, though, Azure is focused on Windows shops and supports lift-and-shift to the cloud for compatible applications. Performance comes as standard, premium and ultra.
Google provides its NFSv3 Cloud Filestore services as a cloud-based NAS for Google Compute Engine and Kubernetes instances, in standard and premium versions.
Microsoft also offers Azure NetApp Files as a premium, enterprise-grade option for Linux and Windows. The supplier says this is aimed at high-performance enterprise applications or web applications, databases and HPC.
NetApp also offers its Cloud Volumes on GCP, with three performance tiers: standard, premium and extreme.
Google’s purchase of Elastifile gives enterprises a further option, with a cloud-based NAS the suppliers say is tailored to enterprise applications.