Architecture Overview
Network File Management and Control lies at the heart of the File Area Network (FAN) reference model, which is currently being defined by the Storage Networking Industry Association (SNIA) FAN taskforce and includes companies such as Brocade, SUN, EMC, Microsoft, IBM and Njini.

Network File Management and Control decouples the end user from the management of their files, in order to reduce complexity, enabling Unstructured Data to be managed more effectively and more efficiently.

Njini’s Network File Control understands Unstructured Data, from its very point of creation, and uses that understanding to make the appropriate decisions required to manage it in a way that makes sense to the business.

At its core the Network File Control architecture is modular and extensible, which enables the system to adapt to the storage infrastructure as it evolves over time. In addition, Network File Control protects legacy investments, by operating transparently across existing file servers, network storage devices and network infrastructure.

Network File Control works by monitoring file operation request (such as open, close and save) by protocols, such as CIFS, and maps these protocol operations to a meta-data database to maintain the expected file system semantics.

Users remain unaware of Network File Control and continue to access their files in the traditional way, their normal workflow remains intact. The “user” may be a person using Microsoft Office on their PC, or a bespoke application that simply creates or move files on the Njini system.

Policy Engine
A policy engine lies at the heart of the Network File Control system and actively manages the user’s Unstructured Data, making real-time policy based decisions on all file operation that are subject to policy. The policy engine is able to support the delivery of a wide range of “file services”, from within the network, by leveraging a variety of meta-data sources. For example: Files Services may permit, deny or modify certain file operations, move certain files to an appropriate tier of storage given their business importance, or simply notify the system administrator of any file operations that attempt to traverse access control permissions.

Meta-Data, Extended Meta-Data & “Helpers”
Network File Control leverages a rich set of meta-data resources, when performing real-time, automated policy decisions: -

• File System meta-data (File name, size, created, modified, last accessed etc.)
• Active Directory® object meta-data (Identity, Role, Location etc.)
• File Content meta-data (Author, Document Title, Footer, Key Words etc.)

Network File Control actively references the underlying file system(s) to gather basic meta-data information about the Unstructured Data files. Whilst this “What” meta-data information (File name, size, created, modified, last accessed etc.) is fundamental, it only provides a small portion of the total information required to construct effective data management policies.

Network File Control also references Active Directory® to gather additional meta-data information relating to the creator or user of a file. This additional “Who” meta-data (Identity, Role, Location etc.) is absolutely crucial, as it provides the individual and organizational context required to construct policies that align Unstructured Data with business requirements.

If required, Network File Control can also leverage “Helpers”. “Helpers” are runtime loadable libraries that may be called upon by the policy engine, to extract additional “extended” meta-data (Author, Document Title, Footer, Key Words etc.) from within the file content itself.

Global Namespace
A virtual file system abstracts the physical nature of the underlying storage infrastructure and may span multiple network storage systems or servers. The virtual file system is the foundation of the native global namespace function, which can also participate within other global namespace schemes, such as Microsoft DFS.

Logical Storage Tiering
The abstraction of the physical storage infrastructure allows Unstructured Data to be dynamically “tiered”, as data may be stored on any network storage device, regardless of its cost, availability or performance characteristics. In addition, logical tiering aligns Unstructured Data with business requirements by separating critical working data from stale data, which significantly reduces the backup window.

Deployment
Network File Control integrates seamlessly with existing backup solutions and is able to leverage common legacy snapshot technologies, as well as more recent industry standard snapshot technologies, such as Microsoft VSS.

Network File Control is deployed as a highly available, clustered system. Both the front-end and back-end network connections are able to leverage existing multi-pathing and fail-over mechanisms, in order to provide further resilience.

Summary
Network File Control allows the automation of data management in a manner that is transparent to the end user, effectively decoupling the end user from the knowledge of the intricacies of the physical storage infrastructure. This enables the storage to be managed by the Njini system, while meeting the objectives for performance, data protection and security within the environment.

Operational issues are addressed by allowing storage technologies to be mixed together to provide common and specialised capabilities that can be provided to users under the guidance and direction of policies.

The architecture is both extensible and scalable and will evolve with the storage infrastructure over time, while providing feedback and reports on the effectiveness of the strategies that are driving that evolvement.