Network-attached storage (NAS) fools servers into thinking they are talking to their own local direct-attached storage. In fact they are talking to a file share that has been allocated to them on a networked storage array. Network Appliance and Auspex started this idea and popularised it with their filers. NetApp has since progressed into SANs and other sophisticated strange ideas. Vendors like BlueArc are making specialised very high performance NAS, and Iomega and others are pumping out commodity NAS boxes.
Who needs anything else in the shared and consolidated networked and shared file storage box area?
We all do apparently, because we don't necessarily want a separate networked file storage device. We want to share files and we want to grow or contract our shared file storage space as our business needs change. That can be hard to do with a single networked filer. But suppose you could link multiple filers together, and add in direct-attached storage as well, and so create a distributed shared file storage construct.
We could call it a logical NAS. It would work by creating virtual networked storage out of existing NAS boxes and direct-attached storage. Servers would still think they were addressing their own local drives. What would actually happen is that they address a file share which could be on another server's direct-attached disks or on one or more NAS devices.
The accessing server's file request goes to a virtualising NAS box which re-directs it to the actual physical storage where the files are held.
Several companies are supplying or developing such technology. They include Rainfinity, Acopia Networks, NeoPath and NuView.
Acopia is a prime example of a distributed, file-based storage aggregator. The company says that its Adaptive Resource Switch (ARX) enables an enterprise to globally present, deliver, adapt, and optimize file storage resources. It preserves existing file storage infrastructure investments. Acopia switches are high-performance scalable devices built to function as in-band NFS and CIFS proxies for client-server file access exchanges.
It aggregates heterogeneous file storage devices and provides a configurable, consolidated Global Namespace or access point to this aggregated storage pool. By independently managing the client presentation, Acopia switches eliminate the need to reconfigure clients during storage provisioning; adds, moves, and changes can be performed seamlessly. This consolidated Global Namespace also allows users to access file storage devices regardless of configuration or location.
ARXs can be interconnected for enterprise-wide scalability. Intelligent data distribution policies allow companies to place data across the corporation to improve performance by locating file data closer to where it is needed. The movement of data is performed without file access disruption or client reconfiguration. Acopia switching products are also able to leverage the enterprise network to automatically replicate data sets off-site to provide more efficient data recovery mechanisms. In the event of a failure, replicated data sets are quickly accessible to client applications using the same client presentation managed by the Acopia switch.
Files/directories can migrate from a storage resource with limited free space to one with more free space-without disrupting applications. Acopia's switching products also enable storage class differentiation and let the enterprise dynamically place data on appropriate classes of storage, significantly improving utilisation by freeing capacity within existing file storage systems.
SO here we have a global name space, in-band virtualised file access, and support for information lifecycle management with tiered storage and transparent data movement. These themes are reiterated by the other vendors.
NeoPath Networks is emerging into the public eye as its technology gets closer to shipping. NeoPath's Network File Management technology is claimed to offer: File server and NAS system growth and consolidation on demand; zero user disruption from changes; Intelligent Unified Namespace for transparent global file access; transparent, policy-driven file migration; tiered storage to supplement expensive NAS servers; and lower CapEx and OpEx.
The company states that, "File Director is a network file switch that lets you create your own directory tree structures out of existing shares and exports, providing a highly customizable global namespace for NFS and CIFS files. It’s a simple, low-cost appliance that doesn’t store any user data, but makes your existing storage much more flexible and economical. Users no longer need to know which particular file server holds their data, and administrators no longer need to be limited by the constraints of existing servers. Instead of unused or underutilized storage trapped in isolated NAS or file server silos, all of your file storage capacity can be used effeciently to meet the cost and performance requirements."
"The File Director is totally transparent to clients and servers. No client or server software agents, drivers, or other shims are installed anywhere. Yet it goes well beyond basic file virtualization. Within a single File Director, you can create flexible file hierarchies that are organized to best meet the needs of different departments within your organization."
NeoPath products are expected to be formally announced pretty soon. Expect them to run on an industry-standard LAN-attached server and cost around $50,000.
NuView has a StorageX product. On the company's web site we learn that it logically aggregates distributed file storage and provides administrators with policies that automate data management functions, including heterogeneous migration, local and remote replication, client failover, capacity optimization, tiered data management, disk-to-disk archival, as well as data classification and reporting. StorageX creates a unified view of data distributed across heterogeneous storage platforms. NuView says it simplifies enterprise file storage management in heterogeneous server/filer environments (Network Appliance, Windows, Windows NAS).
StorageX uses a Global Namespace to unify heterogeneous file storage that is distributed throughout the network. It presents users with a single, logical view of distributed files. A Global Namespace does for file storage what DNS does for networking. It allows clients to access files without knowing their location (just as they access websites without knowing their IP addresses).
StorageX lets administrators aggregate file storage across heterogeneous, geographically distributed storage platforms and view this storage as a centralized resource. This virtualization of the storage architecture can span an entire corporate environment and makes the data’s physical location irrelevant to end users.
The company says that by using its products, the total cost of ownership of that storage is less and users get a much better storage service. Storage/X is a fifth-generation product and can be regarded as an, obviously, shipping product and a mature one.
Where does this leave us?
If you have a messy storage environment with several NAS boxes and direct-attached storage on servers, and if you can't consolidate all or most of this to a single NAS regime, then a virtual NAS may work for you.
Clearly a single NAS unit is simpler to manage. It's just one box and one supplier. But the virtual NAS built out of distributed direct-attached and network-attached storage assets may be a practical alternative. It certainly seems worth aquainting yourself with the technology and becoming familiar with the concepts and its possible applicability to your own circumstances. If nothing else it may be an effective stick to beat down NAS supplier's prices.