Data is critical to businesses of all sizes, and the idea of being able to store it anywhere and access it everywhere is immensely appealing.
This is the promise of global file systems (GFS), which aim to break the link between physical storage and the application and user by aggregating on-premise and cloud data into what is effectively hybrid cloud storage with file (and in some cases, block) access.
Using a GFS, an application or user views files as if they are all in one place even though data may reside in multiple physical locations.
Here, we look at how a GFS does this, and the different forms global file systems can take.
Files, file systems and namespaces
Larger organisations typically have multiple data stores and a range of different file systems.
This creates data silos and makes it harder to share information or move it between applications. Throw the cloud into the mix and it becomes an even bigger challenge. But by moving to a GFS, organisations can take advantage of cloud and hybrid storage.
Global file systems usually build on object storage, which itself uses a global namespace. This allows data to be stored in multiple locations, but appear to an (object storage) application as if it is in one place.
Uptake of object storage is being driven by the hyperscale cloud providers because it allows them to operate at the scale they require across multiple datacentres.
Global file systems take this further, allowing enterprises to distribute data across their local storage and public cloud resources.
Global file system features
A GFS operates a global namespace to aggregate storage in different physical locations. Object storage uses a flat structure rather than hierarchical directories. Each object has a unique identifier that allows systems to scale to vast quantities of data without becoming unwieldy.
However, most applications cannot access object storage directly. A global file system addresses this by linking its file service – usually in the cloud – to local hardware or a virtualised application on the user’s LAN.
This hardware or virtual machine (VM) acts as a cache and, critically, ensures application compatibility. But all the data remains in a single namespace to allow data storage across cloud resources, private clouds and local arrays in the on-premise datacentres.
Global file systems also provide file locking that prevents multiple users from trying to write to files simultaneously and causing data errors. Object storage does not have file locking. File systems do, so GFS suppliers add this to provide access in the way applications expect.
Suppliers add other features based on their position in the market. These include migration services that can help firms move data from legacy network-attached storage (NAS) to the global file system. Support for various operating systems and applications that allow customers to access storage without refactoring applications is also provided.
Use cases for global file systems
The main use cases for global file systems centre on the need to store large and growing volumes of unstructured data.
A GFS works where organisations want the benefits of a global namespace and the scale of the cloud, but are not in a position to move directly to object storage
In this, their application is very similar to object storage more generally. But a global file system provides more control over where data is stored – with the option to keep some data on-premise – and compatibility with existing operating systems and applications.
A GFS works where organisations want the benefits of a global namespace and the scale of the cloud, but are not in a position to move directly to object storage. It also works for organisations that have significant investment in on-premise storage and want to continue to use it.
Global file system suppliers and what they offer
Each GFS supplier has a slightly different approach, and even a different name for its products.
CTERA uses Edge, a caching filer; CTERA Drive, an agent for endpoint devices; and VDI for virtual workspaces. CTERA says it has no capacity limits, and provides file-sharing services and device-level access from Windows, Mac, Linux, Android and iOS.
Hammerspace offers a global data environment that combines the cloud, datacentres and the edge into one dataset. This then provides access, data tiering, file versioning, deduplication, snapshots and antivirus. Hammerspace can also access cloud, datacentre and legacy DAS, SAN and NAS filestores.
Nasuni positions its UniFS as the building block for its File Data Platform. On-premise edge instances provide local cached access, file management and orchestration. Nasuni IQ adds performance analytics and application programming interfaces (APIs).
Panzura positions itself as a single platform for unstructured data. Its global file system is CloudFS, which it says creates a single, optimised dataset for enterprises. The service also has direct support for mobile devices, which sync directly to the cloud. Panzura also offers cloud block storage.
Peer Software’s global file service offers hybrid and multicloud support. PeerGFS is software only, with active-active sync, a global namespace using Microsoft DFSN (Distributed File System Namespaces) and object storage integration. It also supports cloud backup and replication.
Qumulo’s software-defined file system covers on-premise and cloud storage, and targets high-performance and high-capacity applications.