Network file virtualization.

Virtualization is key to managing today's demanding file storage requirements. One of the most promising--and challenging--developments is network file virtualization (NFV) that enables unstructured data management without disrupting end-user or application access. Done correctly, network file virtualization optimizes networked attached storage (NAS), eases storage management overhead, simplifies end-user access, and enables additional storage management functionality.

Strategic approaches like ILM. Storage Grid and Utility Computing require data to move freely across the environment without disrupting end-user or application access. Standards-based NFV offers this required non-disruptive data movement across heterogeneous environments, which represents a significant change in the ability to effectively manage NAS environments.

Government and industry regulations have placed new demands on the storage infrastructure right in the middle of flat IT headcount budgets and unprecedented storage growth. Strategic Research Corp. recently published a three-year forecast predicting a seven-times increase in storage capacity but no associated increase in IT personnel to manage the larger pool.

File virtualization technologies like NFV were developed to simplify and automate NAS and unstructured data management while enabling client and application data access.

NFV Benefits

According to IDC, storage administrators must increase their efficiency by 60% just to stay even with the anticipated growth in storage capacity. NFV not only helps administrators meet the storage management challenges of today, NFV drives tremendous benefits.

NFV Components

NFV is based on a number of widely deployed and mature technologies integrated with some newly emerging technologies. Mature technologies include file servers, file systems including recent innovations for clustering, distributing and spanning the wide area networks; along with data protection utilities including snapshot and backup, and global names spaces including DFS and Automount.

However, important functionality has traditionally been missing:

* Active data cannot change physical location due to disruption of user and application access

* Global namespace presentation and management is too complex

* While Automount is prevalent, it has limitations and is not part of the NFS protocol standard

* Identifying and resolving capacity and performance management issues is too manual and labor intensive

* Staging data between online and near line storage is not automated

* Consolidations are continually delayed so the impact on data availability and user disruption are IT labor-intensive and error-prone

The emerging technologies enabling NFV include file management applications, transparent file protocol switching, standards-based global namespaces, and global namespace management.

File Management Applications

NFV enables file management applications that transparently manage active data. Purpose-built applications simplify management tasks and automate file management by identifying capacity and performance issues and then directly resolving those issues. File management applications go well beyond existing tools that provide monitoring capabilities and storage status information. File management applications, with NFV as an underlying foundation, allow administrators to optimize storage instead of simply checking status.

These essential, NFV-enabled network storage applications include:

* Capacity management. Automatically identifies over-allocations at the file server, volume and directory level; and takes corrective action.

* Tiered storage management. Automatically stages data between high-speed and near-line storage; transparently supports service levels without disrupting access.

* Performance management. Automatically identifies performance bottlenecks at the file server, volume, and directory level; and optimizes I/O.

* Consolidation. Pools multiple file servers to appear as one. Can be permanent or temporary as part of consolidations.

* Disaster recovery. Synchronous and asynchronous content distribution across multiple locations.

* Regulatory compliance. Automates regulatory compliance data retention, retrieval and authentication.

Transparent File Protocol Switching

The key to NFV is the ability to manage storage without any user or application service disruption. The batch window continues to shrink while the amount of storage to be managed continues to grow. Management operations must execute without impacting data availability.

Users and applications cannot cope with a file's physical location changing while actively reading and writing data. Transparent File Protocol Switching on the network is therefore required to manage active data.

The key to Transparent File Protocol Switching is the ability to provide location transparency for data actively being accessed without introducing data integrity, data access or performance risks. True switching technology removes the data integrity and access risk by routing traffic, rather than storing persistent data--even file system or global namespace related data.

To virtualize a large number of file servers, filtering is critical to switch only the required traffic and bypass processing all other traffic. Transparent file protocol switching requires the examination of only a small fraction of the network traffic. A proper filtering architecture allows file protocol switching to "front-end" literally billions of files and hundreds of thousands of IOPS.

Global Namespace

File Virtualization also requires a global namespace. A global namespace provides a centrally managed logical name that abstracts the physical location. DFS is part of the CIFS standards servicing the Windows environment. Automount is a de-facto standard supporting Unix/Linux environments. Both of these technologies provide out-of-band lookup and scale well by not impacting I/O operations. Both technologies have an unfair advantage by already having their client components present on Windows and Unix/Linux clients.

Enhancements to the presentation and management of these namespaces are required. IETF is working on a single mount point namespace to be delivered as part of the NFS v4 standard.

While NFV requires a global namespace, it does not require a separate file system that resides in the network. A virtualization approach that places a proprietary namespace in the data path does not scale. Given the high number of files in a typical enterprise, a bundled global namespace and file system will not be able to attain global levels of scalability--there are simply too many files.

NFV leverages industry standard namespaces to perform directory look-up services, thus avoiding any requirements to load proprietary client software. Clients interrogate a namespace server to resolve the physical data location without incurring the limitations of routing I/O operations through a namespace server.

Global namespaces are an essential component of NFV. Using industry-standard global namespaces provides a safer, more scalable, and easier to deploy alternative than proprietary always-in-band approaches. Certainly, adoption of standards-based namespaces will outpace and ultimately supersede any proprietary in-band namespace solutions.

NFV: Changing Storage Management

NFV is changing network storage management by easing management overhead, increasing utilization, decreasing over-provisioning, resolving performance bottlenecks and lowering TCO. NFV allows administrators to support heterogeneous storage environments and to optimize networked storage across different vendor platforms and storage tiers.

NFV architectures support distributed enterprise-wide file access across multiple protocols, multiple operating systems and multi-vendor storage systems. NFV should therefore, be standards-based and operate in multi-vendor environments without needing a proprietary file system, file protocol switch or global namespace.

In traditional file server environments, adding data storage greatly increases management burdens and data exposure, and impacts performance and availability. NFV allows administrators across large-scale, multi-vendor storage environments to efficiently manage their NAS networks regardless of their size, regulatory requirements, data volumes, service levels, or high availability requirements.

Storage Management Challenges         NFB Benefit Examples

CAPACITY ISSUES                       INCREASED UTILIZATION
* Average itilization is 35-50%       Large ISP increased storage
* Management per TB is not improving  utilization to 90% and is able to
* Over-provision is too costly        respond to capacity issues without
                                      disrupting end-user data access.

PERFORMANCE ISSUES                    PRODUCTIVITY IMPACTS
* User productivity impacted by poor  One of the largest US financial
  response time                       services firms achieved over a
* Application throughput limited by   half million in labor savings
  IO bandwidth                        alone through network file
* New devices don't address           virtualization
  bottlenecks

TIERED STORAGE                        CAPEX SAVINGS
* Nearline storage represents huge    A major semiconductor company
  capital expenditure advantage over  reduced storage CAPEX by 50%
  online                              through the greater use of SATA
* Limitation to dynamically manage    enabled by NFV.
  data between online and nearline

STORAGE CONSOLIDATION                 BUSINESS EFFICIENCY
* Consolidation projects slip         Fortune 500 firm completed a 30TB
  because of the organizational       consolidation project in 1/10th
  impacts                             the extimated time with a 4 week
* Consolidations require data         payback.
  relocation along with security and
  access settings

John Schroeder is president and CEO of Rainfinity (San Jose, CA)

www.rainfinity.com