Service provider takeaway: Storage service providers can turn to NAS-based file virtualization to solve the problem of unstructured data growth at client sites.
The next frontier in virtualization is upon us; we have the technology, and customers have the need. Are you ready to deliver? We've already seen the absolute dominance of server virtualization by VMware. We have seen companies like Compellent and 3PAR do what many thought was impossible -- carve out a share of the primary storage market through their storage virtualization and thin provisioning capabilities. The next frontier in storage virtualization is file virtualization on network-attached storage (NAS) systems.
NAS-based file virtualization promises to be a good play for the channel. It gives service providers an opportunity to pitch customers on a way to solve real file management problems, with immediate ROI. Implementing the technology also generally requires integration work and professional services, which service providers should be happy to supply. Storage service providers can even use the opportunity to offer complementary storage equipment along with those services.
In the 1990s, one of the more common conversations was how to protect databases. Oracle and other databases were experiencing a big ramp-up, and keeping them protected was critical. Channel partners that could demonstrate an expertise in database protection were bringing value to customers and building a profitable business. While the database protection continues to be a productive market, customers have begun focusing on managing unstructured data.
Unstructured data consists of common files, like office productivity reports, images or scans of documents that land on NAS or file servers without being part of a database. This proliferation of data files is problematic for customers because hard copies are becoming less common, but legal and regulatory standards require businesses to keep information on hand, forcing data to be stored electronically. Unstructured data is filling up expensive primary storage and increasing backup windows. Resellers that develop a practice around managing this data, moving it to less expensive storage devices and decreasing backup resource requirements, are in a good position to create a viable service for customers with a strong return on investment.
How can you manage and move that data to secondary tiers?
A viable answer to this question is file virtualization. There are several methods for file virtualization, typically either doing it via software, by the operating system (like Microsoft DFS) or with a purpose-built appliance. Since much of this unstructured data is residing on NAS storage, I think a service provider is better off focusing on an appliance. Most NAS storage is also appliance-based and as such can be difficult if not impossible to install software on. A file virtualization appliance does not need to install software on the NAS units and sits inline between the users and the NAS. Think of the NAS virtualization appliance as a DNS server for files. In the DNS world, most of us do not know the IP address for Yahoo.com; we just know that when we type "Yahoo.com" in the address bar, it appears in our Web browsers. The same mechanism is at work in file virtualization; rather than needing to know where a file is located, a user simply requests the file to open it on their computer.
Once freed from having file data tied to specific locations, many tasks become much easier to perform, such as migrating data from an old file server to a new one in real time or moving older data to a certain class of storage, such as a new deduplication archive appliance that further reduces storage costs. NAS virtualization can also help you manage compliant data that is governed by legal or government regulations by providing the tools to identify and automatically move it to a secured, WORM (write once, read many) or encrypted storage platform.
The single biggest advantage of file virtualization is that it eliminates stub files -- placeholders that are left behind by traditional file management applications when a file is moved, pointing to its new location. Stub files aren't an ideal mechanism because they need to be backed up and stored, and problems ensue when users inadvertently delete them. In many cases, stub files simply don't work properly; they don't point to the right location. NAS virtualization tools store all the metadata on the NAS virtualization appliance and do not need to leave stub files, making the use and protection of the file management environment significantly easier.
Beyond the stub file difficulty, another problem with most traditional file management tools is that they are out of band. This means the software has to walk or scan the file systems to perform operations on the files that they contain. Even though this does not require IT personnel's particular involvement, this process can be very time-consuming; scanning servers with millions and millions of files can take weeks in some cases. To improve performance, an alternative to the file system scan that other traditional file management applications employ is to install an agent on every file server, which can conflict with that server's OS operations and has been proven to create instability issues in the past.
Most file virtualization solutions are in band and operate on file data in real time as it travels through the appliance so there is no need for stub files or agents.
Components and professional services
To implement a file virtualization system for customers, you need the appliance itself. There are several out there, most from companies with strong channel programs. I would consider Attune Systems and F5 Networks (F5's recent acquisition of Acopia put them in this market). In addition to the appliance itself, it will likely make sense to offer a lower tier of storage to move data to, and possibly a tier of storage that can handle specific retention requirements. I strongly suggest including a disk-based tier that can perform data deduplication, since the ability to identify 5 TB of old data and move it to a device that can store that in less than 1 TB will be very compelling to customers.
While implementing a NAS virtualization appliance can be as simple as installing the appliance and globally tagging data for movement based on age or type, a reseller should be able to offer professional services to make the system a solution. Since a NAS virtualization appliance is an inline device, integration requires planning and experience to graft it into the environment. A professional services deliverable should be created to help a customer move to this environment, plan where file are placed, choose the best class of storage, and lastly help develop rules and policies to enforce based on the new capabilities of file virtualization.
If you are thinking about expanding into file virtualization services, consider the following: The market is not yet overcrowded with channel partners that are focused on it. You have the opportunity to offer a complete solution that includes the appliance, storage hardware and services that can add value to your customers and your business.
About the author: George Crump has 25 years of experience in the storage industry. He has witnessed the birth of enterprise backup systems, network-attached storage and storage area networking. Prior to founding Storage Switzerland (http://web.mac.com/georgeacrump), George was chief technology officer at a major storage reseller where he ran a series of interoperability labs that tested and launched new products for the organization.
