Something’s missing from global file systems

Analysis: Rarely does a CIO say their business has a global file data access problem and they need a supercloud-type global file system to solve it. Yet, arguably, they do and, in principle, a global file system operating across a hybrid and multi-cloud environment is exactly what they need to fix it.

A global file system is one that presents each user with the entire organization’s set of files, folders and directories as a single entity – even though the files and folders may be physically stored on different vendors’ filers and in different public clouds, in different regions and instance types within those clouds.

A supercloud, according to Wikibon, is “a set of services abstracted from the underlying primitives of hyperscale clouds (compute, storage, networking, security, and other native resources) to create a global system spanning more than one cloud.” The storage part of that is what we mean with the global file system idea.

Global file systems exist within a number of vendors’ offerings. For example, NetApp has its data fabric with the ONTAP file system and services running on premises and in the public clouds. Applications and users accessing ONTAP files can use them anywhere there is an ONTAP filer presence. As we said in June: “Customers can move applications using ONTAP to, from, and between their datacenters and the public clouds, and find the familiar ONTAP environment in each.”

Other supplier-specific global file access environments come from CTERA, Egnyte, Nasuni, and Panzura. This group of suppliers started out as general file synchronization-and-share and collaboration suppliers, delivering remote access to files stored in the public cloud through cloud storage gateways. This original intent – file sync-and-share – was not congruent with NetApp’s Data Fabric concept.

We wrote four years ago that NetApp’s Data Fabric “aims to deliver a single, consistent IT environment in which workloads and applications can move between and use mainstream public clouds and customer on-premises datacenters.”

The four sync-and-sharers have implemented aspects of this by building file-based collaborative data services with local caching. You could replicate this with NetApp’s ONTAP but would need a filer in each location and additional software – a collaboration-focused presentation layer. Evaluator Group consultants say both ONTAP and the cached cloud-based global file systems are capable of providing on-premises file access, performance, and familiarity with an attached cloud component, but they have fundamentally different designs and trade-offs.

Komprise, with its Intelligent Data Management platform and data movement capabilities, focuses on unstructured data management activities and presents itself as a service to be used by customers along with NAS filers, object stores, and collaborative file sharing service suppliers – not a replacement for them.

The Nasunis and Panzuras will say that they can replace on-premises and in-cloud filers. Not so, says Komprise. It moves files and objects around an existing file object global system estate to optimize overall fie storage and management. 

Komprise answers a file management and storage problem. The sync-and-sharers answer a file access problem for remote users. The hybrid filer suppliers, like NetApp, Qumulo and Dell (PowerScale), answer a file access and lower cost long-term (cloud-based) file storage problem.

Quantum answers a media and entertainment workflow problem with its StorNext set of products supporting distributed and remote workers creating and building complex special effects and video sequences for movies. This, in a way, is a highly vertical market-specific version of general file-based collaborative work support by CTERA, Egnyte, Nasuni and Panzura, with API-level integration into dedicated applications in the workflow environment.

No one supplier in these four groups of suppliers is answering a general and clear global file system need. Not one built their product in response to that overall need – they built them in response to aspects of it, and their offerings partially overlap.

Enter Hammerspace. It has devised a global file system and services in response to its founder’s view that this is what the world needs, not point products to fix point problems. But Hammerspace can’t make its major sales pitch that it fixes global file system problems because customers aren’t generally buying it for this purpose. They don’t perceive they have a problem in those terms.  

They have specific needs – such as a fast file/object storage on-premises system with the ability to have a cloud-based backend for cool and cold data storage. This could be a situation faced by a Pure Storage sales team or partner. FlashBlade fits the first need, but they need a component to fix the second need. Panasas could be selling its storage into an HPC customer and need the same capability. There’s a natural field-level partnering opportunity here between these suppliers and Hammerspace.

So we have five kinds of suppliers, all punting products to fix specific problems which all represent examples of global file system products – but with their different focus. As their competition continues and perhaps intensifies, will the suppliers make their products more similar?

Will Dell, NetApp and Qumulo add remote work collaboration features to compete more effectively against CTERA, Egnyte, Nasuni and Panzura? Will these four add features to strengthen their entertainment and media video production market appeal? Will Quantum add more general collaboration and lifecycle management features? Could Komprise add more data services to compete with the others? Will Hammerspace add more infrastructure services so partners and customers can use it to compete better with the other groups of suppliers?

It all depends on whether the suppliers perceive that they are competing in a single global file systems market, and whether buying customers want products and services that give them a general global file system capability. We think the answers are: first, some do and some don’t; and second, not yet – but they might.

It doesn’t matter how valid a company founder’s view is of their technology vision, nor how architecturally clean or conceptually elegant a particular product is. Ultimately the market – meaning customers – will decide, and that means specific problem fixes, features, price, performance, etc. All we can do is point out possibilities, watch, wait and see what happens.