The FAN has Started Spinning

Monsoon has already hit the Indian shores and so, who needs a fan now. This

is what I thought when I first came across File Area Network, or FAN for short.

I thought of it as yet another marketing-driven initiative; yet another bottle

to sell the same old wine. I kept on overlooking it for almost a year, but time

and again, it came back chasing me and as an even bigger lump of mass.

Therefore, I decided to bust it right here, before it comes to me again and

rolls over me. If you are also one of those who are being perplexed by this

jargon, please be a part of this busting. We'll do it together.

Need for file management



There is so much of unstructured data, generated in any enterprise today,

that traditional approaches to managing, storing and retrieving them have

started hitting the wall. Ask any data center manager and he'll say that file

management is one of the top priorities for him to look after and that it needs

an immediate panacea. On the other hand, we are also seeing concerted efforts by

storage vendors and innovators to drive innovations around management and

control of files.

In the recent past, we've heard about a wide range of new technologies coming

up in this area, including Wide Area File Services (WAFS), WAN optimization and

application acceleration, distributed and clustered file systems, file

virtualization, file or document management software, file classification

software and file data placement and movement controls. So, what all has changed

to make the storage managers look for file-level control over storage and where

and how FAN fits into the story?

Quite simply, it's the relative criticality of file data that has changed

with time. File-based data is increasingly becoming important, as nearly all

workflow processes today run through a file infrastructure. The growing

importance of data contained in files owes to behavioral changes that have taken

place with time, like we are now-a-days, closing deals on e-mails and minutes of

a Board meeting are being recorded in MP3 format, which are not structured, yet

we are required to keep them. Sometimes for regulatory reasons and on other

occasions for their potential value in future litigation.

Today in an organization, we have unstructured or semi-structured data stored

in these files, as a large portion of its total storage. This data has a

potential to grow at a much rapid rate than structured data. Besides, there is

growing complexity in terms of mixed vendors, platforms and file systems. The

application demands and availability requirements have also increased. This all

simply demands the deployment of advanced file management services.

Sense of déjà vu



With traditional storage, we see a tight coupling of applications with

storage, which has so far been preventing the evolution of storage management

solutions. A FAN unknots these tight bindings and enables advanced storage

management services. It is analogous to SAN, as both talk of a unified pool of

storage resources.

The difference lies only in the fact that it talks of abstracting and

extending the 'area network' concept to a higher layer of the infrastructure,

that is, a file system. SAN replaced physical block storage connections with

logical connections, while FAN replaces physical path names with logical path

names. That means FAN could turn out to be a similar revolution for files, or

may be objects, as SAN was for the block storage.

What's a FAN?



The goal of a FAN is to provide an enterprise-wide intelligent platform for

cost-effective delivery of file information with a better level of file control.

FAN is an umbrella term which refers to an architectural model encompassing

storage devices, file-serving devices and interfaces, namespaces, file

management and control services, end clients and connectivity.

A FAN offers pervasive controls of all file-based information. For this, the

infrastructure maintains an inventory of metadata and content values. This in

turn helps in launching a process as ambitious as Information Lifecycle

Management (ILM), in a true manner.

Regardless of the physical device, it offers file visibility to the user

applications. Intelligent search on unstructured data is possible in case of

FANs. Then, non-disruptive migration, replication and de-duplication of data are

possible features which can be worked out on top of a FAN foundation. This will

result in measurable RoI for file management.

Finally, with all this consolidation, we will also have improved space

utilization and highly improved service levels.

The heart of a FAN



The way a file system object is kept and retrieved in a FAN framework is

referred to as the object's namespace. This is the heart of a FAN. Today, we

find three kinds of namespaces in a FAN infrastructure. Most enterprises use a

combination of these to address the issues involved.

1. Non-shared namespaces: This is a user-level presentation of information

corresponding to a file system image belonging to a particular physical machine.

Sharing of information across multiple file system images is not possible in

this case.

2. Shared namespaces: These are platform-specific and not intended for

deployment across all end clients in an enterprise, rather they are targeted at

a subset of this volume. They result in file visibility, collaboration, and

performance issues for this subset of client devices/applications. Clustered NAS

environments and clustered or distributed file system deployments are an example

of such a kind of namespaces.

3. Global unified namespaces (GUN): They enable a true heterogeneous FAN.

They offer a complete abstraction of all file-level information present in an

enterprise. Once you have a unified namespaces well-defined, you can think of a

significantly improved management control over storage and also will be able to

leverage it in the best possible manner.

FAN services



These are some basic services which we expect a FAN to deliver, apart from

advanced services which it will provide with time. The basic set of services

includes: a global unified namespace to organize storage in an overlay

namespace; migration of files from one server to another; moving files via

policy to the right storage at a point in time (ILM), moving files to better

distribute load or capacity; and replicate files to support any kind of

failover.

A FAN supporting all of the above services is very useful. It supports the

day-to-day cause of a storage administrator in an enterprise. Apart from these,

a FAN can also enable a wide-range of advanced services that will take storage

management to new levels. These advanced services include data classification

and optimized placement of data, application acceleration, and access control

and auditing. The Information Classification and Management (ICM) software,

enables content-level indexing of all information contained in a file

infrastructure, which in turn supports policy-based controls, access and

retention. This optimizes the alignment between storage characteristics and

business needs. Optimal data placement ensures appropriate performance and

utilization levels for servers and storage. This can either be achieved through

some in-band network resident approach, like Network File Management (NFM) and

ICM, or through a distributed software approach.

Application acceleration is another advanced service which is designed to

optimize the performance of critical applications. As FAN extends through

geographies, it's necessary for it to support wide-area connectivity into its

namespaces. There should not only be a connection, but a connection as fast as a

LAN. Various WAN optimization technologies along with WAFS are making this a

possibility.

State of the art



The concept of File Area Networks is still in its infancy and will continue

to evolve in coming times. It took SANs almost seven years to settle down and

become as effective and common, as they were initially planned to be. Similar is

the case for FANs, though they may take a little less time to become popular.

For the time being, IT managers, who are seriously viewing at investing in a

file networking based solution, need to seriously weigh the options available

and study their pros and cons. Thankfully, today there are a bunch of solutions

available and the good news is that tier one vendors such as Brocade, EMC, HP,

Microsoft and NetApp are all lined up with FAN-based solutions. It seems as if

the FAN has well and truly started to spin.