Global namespace: The future of distributed file server management, part 1 - Storage Networking

Computer Technology Review, Feb, 2003 by Rahul Mehta

Global namespace is the key to effective, efficient management of distributed file storage -- it does for file storage what DNS does for networking. A namespace allows clients to access files without knowing their location (just as they access websites without knowing their IP addresses). A namespace also enables an administrator to aggregate file storage across heterogeneous, geographically distributed storage devices and to view and manage it as a single file system.

Introduction

Today's file systems were designed to take advantage of the storage architecture of the 1960s, which was far less distributed and characterized by static links between clients and storage volumes. In the 1990s, however, the world of file storage became distributed, making file system management more complex and costly. What is needed is a new approach to file management, one that simplifies file management for users as well as administrators.

File Management Challenges

The explosive growth of data storage and massive proliferation of file servers and NAS appliances has created a management nightmare for company data centers and storage administrators. Every server and filer is a storage island and an independent file system that requires individual management on a regular basis. Companies are looking for ways to simplify file system management and reduce storage management costs.

Corporate users and applications find it difficult to navigate today's file systems--they must map (or mount, in the UNIX world) the shares (or exports) they access while carrying out their tasks. (Figure 1) User issues with the traditional file system paradigm include the following:

* Users must know where files are located

* Users must map to multiple volumes in order to access required data

* Cross-volume data searches are difficult

* This configuration is not highly reliable, available, or scalable

* There are many single points of failure throughout the distributed file system

* If files are moved or storage reconfigured, user access may be interrupted, as their shortcuts and login scripts must be modified to access files in their new location

For administrators, the issues are even greater. Administrators face significant challenges as they work to provide efficient file storage management. Challenges that drive up hardware and administrative costs include tasks that administrators face on a daily basis:

Adding file servers. Adding the first server (whether NAS filer, DAS, or server connected to a SAN) is easy -- it contains its own file system and is easily implemented in the network. When a second NAS box is added, the administrator must once again set up network shares and inform users of its existence so that they can mount/map to it. Each successive NAS addition requires redundant administrative setup and provides additional complexity for administrators and users.

Load balancing storage and migrating data. If an administrator determines that one file system is 100 percent utilized and another is underutilized, he or she cannot move data from one to the other without affecting users.

Preparing for disaster recovery. Administrators are tasked with making file systems highly available and recoverable without interrupting business, which is exceedingly difficult in a distributed environment.

Reconfiguring file systems. Users are aware of the name and location of the servers on which their data resides, as they must attach to each machine in order to access files. This complicates the administrator's job, as be or she cannot add, move, migrate, or rebalance storage without interrupting users' access to it. In addition, all such changes require some reconfiguration to the client machine. The hard dependencies between users and physical storage makes effective, efficient file system management an impossible dream in today's environment.

The Answer: Global Namespace

A namespace is a logical layer that is inserted between clients (users and applications) and file systems. It provides a method of viewing and accessing files that is independent of the physical file locations. This is a powerful concept, as it means an administrator can use a namespace to logically arrange and present data to users, irrespective of where the data is located. It also gives administrators the ability to add, change, move, and reconfigure physical file storage without affecting how users view and access it.

The goals of a namespace are to: 1) shield users from the complexities of the storage architecture, and 2) enable the administrator to manage the physical layer without affecting how users access files. A namespace is a means of "pooling" multiple file systems into a single, global file system. Ideally, a global namespace can pool storage from multiple, heterogeneous storage types (DAS, SAN, or NAS), and across different storage platforms (Windows and UNIX).

With a global namespace in place, the administrator is able to distribute files in a way that achieves best performance and capacity utilization, and clients access them via the logical namespace. When storage is added or consolidated, and files are moved or renamed, clients are automatically redirected to the files in their new location without ever having to know (or care) that they were moved.