Managing data at the edge: taming the 'wild frontier' of remote or branch office data.

One of the last things an IT administrator wants to hear is that "the network" is running slow. Seen as a barometer of the overall quality of IT services, fast access to application and file data has become such a user requirement in most enterprise organizations that even a few seconds' delay in opening a file or application can be enough to generate new calls into support.

Thankfully, high-speed, high-availability networks and fast-moving networked storage architectures tend to minimize the risk of excessive user access delays for today's corporate data center. Likewise, initiatives to consolidate and centralize server and storage resources have resulted in not just better access to files and applications, but also a better way for IT to centrally control, monitor, secure and protect the data under its care.

The evolution of networked storage architectures capable of transmitting data quickly in a core/edge topology has done much to hasten speed of application data delivery out to even the far corners of the corporate network.

As data centers continue to mature in their adoption of best practices for centralized, controlled data delivery within the walls of their corporate headquarters, their efforts have begun to turn to the problem child of many IT environments: The growing league of remote, geographically dispersed offices that now represent the data center's ever-widening "edge."

Problems at the Wild Frontier

In a recent Storage World Conference session on remote branch office challenges, Taneja Group founder Arun Taneja gave some indication of the scope of the problem now facing IT. U.S. businesses currently boast as many as 2,000 remote offices, he said, with 75 percent of employees working outside of their company's corporate headquarters, and over 60 percent of company data currently generated outside the organization's main offices.

No small issue, near-instant branch office access to file and application data is a dilemma met by some frustration on the part of IT organizations. Most efforts to resolve the issue tend to come with a healthy dose of trade-offs.

A commonly adopted approach, and one that's often among the more costly, is to deploy separate servers and storage at each remote office in order to serve that office's data access needs. While this satisfies remote users' short-term demands for quick access to data, this approach quickly increases IT's acquisition and management costs associated with each remote office. Since many of these locations do not boast a dedicated, local IT presence, this approach also falls short when it comes to achieving adequate, on-going backup, maintenance and updates of data and applications residing on each remote server.

Central IT policy regarding the secure access, management and protection and auditing of the data that is now taken for granted at corporate headquarters is also hard to achieve at each remote site. Then, there's the issue of remote workers needing to open, edit and save data within one of the same files their peers access as well from within the main offices.

Collaboration and the need to share files across geographic boundaries is an added wrinkle which has led many organizations to institute periodic replication of primary data out to each branch office server. Depending on file/data usage and access patterns at both the remote or central office, this approach can work. But, it also serves to increase the cost of supporting the branch, in terms of the added replication software cost, and the on-going costs to lease or develop sufficient WAN bandwidth required to transmit data changes between the sites.

Even then, confusion regarding different versions of files in circulation may crop up as updates or file saves are made locally by various users within different regions.

WAFS and WAN Accelerators: New Approaches to the Branch Office Conundrum 

Technology innovation is usually not far behind the latest problem to face IT. In this case, innovation and inspiration for resolving the branch office data dilemma came from an unusual quarter. Instead of following conventional wisdom to add infrastructure at the remote office, developers of what the Taneja Group now calls Wide Area Data Services (WDS) appliances began to explore ways that remote offices could still gain acceptable WAN throughput and latency in order to remotely access files and applications located at a company's main office.

Not an easy technical hurdle to overcome, file system protocols like CIFS and NFS were not known to travel well across a WAN. Similarly, even the remote performance of basic edits, opens or save functions in a Microsoft Office-type application could result in extreme latency as hundreds of separate handshakes traversed the WAN, back and forth, from the remote user of a file to the file's central location.

Yet, inventors persevered and soon, two distinct types of technology began to emerge within the WDS arena. These came to be known as wide area file services (WAFS) and WAN accelerators.

WAFS devices aimed most of their technology at the file system layer where they streamlined the handling of file system protocols like CIFS and NFS across the WAN, along with performing functions like remote data caching of files to speed their access by remote users. With a core WAFS appliance typically installed at the data center, and a smaller edge appliance at each office, the WAFS approach offered LAN-like read-write access to shared files that soon began to gain the attention of industry analysts.

"While there are ways to tackle the remote data management problem using traditional replication and remote backup technologies, ESG believes wide area file services (WAFS), otherwise known as remote file services, has clearly emerged as a front-runner technology in this battle," claimed Heidi Biggar, an Enterprise Strategy Group analyst in a January 2006 ESG information management brief.

Performance metrics soon demonstrated that a WAFS solution--like that offered by Brocade Tapestry WAFS, Cisco Systems and others-had the potential to reduce remote access times to a centrally located, multi-megabyte Microsoft Word or CAD file from several minutes down to just a few seconds. The viability of this approach piqued the interest of IT organizations wanting to maintain central control of file data while still meeting the needs of their remote users.

At around the same time, innovative technology from another direction--WAN acceleration-also began to appear on the market. Geared more toward improving and simplifying the TCP transport layer involved in transmitting frequent blocks of application data across the WAN, WAN accelerators also received their share of industry fanfare.

Finally, central IT organizations could see the taming of the remote office frontier clearly in their sights. It wasn't long, however, before the lines between the two technologies began to blur as new versions were introduced appearing to contain some WAFS and WAN acceleration features mixed together. While enterprise organizations were interested in the technologies, it was tough for them to distinguish the difference between various offerings on the market.

Evaluating the Options

While this article is not intended to identify all unique differentiators of solutions in this market, it can highlight a few questions to ask when evaluating the various options:

What end goal are you trying to accomplish: better data consolidation and file management or simply better WAN network optimization?

If data consolidation or file management is your motive, a more WAFS-centric solution may be a better fit. More WAFS-centric solutions usually perform a wider range of file-level functions than their WAN optimization counterparts. An example might include the initial transport of file-related header and footer data across the WAN so that the associated application can be opened right away on the remote users' workstation, as opposed to being left hanging until the final TCP/IP packet is transmitted.

What level of Microsoft integration does the solution offer, in terms of support for both current and future versions of CIFS?

Since Microsoft recently decided to "close" the previously open CIFS protocol, this is an important question to ask. When it comes to future support for new, Microsoft-proprietary CIFS calls. Linux-based WDS solutions may lag behind in their efforts to reverse engineer and support the new calls. In some instances, this scenario can make remote file access either completely non-functional for specific files no longer able to be recognized by the WDS device, or at the least, significantly slower.

How does the solution handle WAN disruptions in order to prevent data corruption, or allow files to still be saved or opened remotely during a WAN outage? What type of safeguards are in place for preventing separate versions or copies of files from being created during a WAN outage or disruption?

If security is important in supporting remote access to Microsoft Office files and applications, look closely at the support the solution might offer for native Microsoft security features like SMB packet signing.

SMB packet signing is a security feature that requires a native Microsoft handshake in order to ensure the user requesting the file is the same one who reads it. If the WDS device in use is not Windows-based, this feature is not supported and must be manually turned off before file optimization can work. In future Microsoft Windows versions, it's also likely that SMB packet signing functionality may need to be turned off by an administrator on a file-by-file basis.

Make Way for the 'FAN'

In the end, most analysts state that a good understanding of your environment's unique needs and user access patterns is still the best way to start when choosing a specific WDS solution. Some, like analyst Brad O'Neill in the Taneja Group, even suggest taking another step back from point technology decisions and looking, instead, at the development of a new file area network (FAN) paradigm. The FAN, he claims, has the promise to offer users universal access to relevant files--from wherever the user happens to be, to whatever storage platform or geographic locale the file happens to be stored on at a given point time.

According to one of O'Neill's recent industry commentaries on FANs, an interesting FAN development is the inclusion of a host of related file services--from namespace functionality, to print services, Exchange Services, DNS/DHCP and SMS--which have already begun to appear in the latest breed of WDS devices, and from early FAN industry leaders like Brocade, EMC, Hewlett-Packard, Microsoft and Network Appliance.

Such emerging functionality may also bear a closer look. A fully functional FAN can loosen some of the traditional boundaries surrounding file management, while further reducing the need for duplicate IT infrastructure and local IT support at the branch office. Plus, it makes better use of each edge device already installed.

Michael Schmitt is a Product Marketing Manager at Brocade Corporation. He can be reached [e-mail].

www.brocade.com