IBM®
Skip to main content
    United States [change]    Terms of use
 
 
 
    Home    Products    Services & solutions    Support & downloads    My account    
IBM Research

Storage Systems - Projects - Glamour

IBM Almaden Research Center


Overview

Glamour is a file virtualization and data management solution that leverages open standards (e.g., NFS) for file access. Glamour provides a common namespace so clients can see a consistent view across a group of distributed, heterogeneous wide-area distributed fileservers. While Glamour is not a globally distributed file system, it enables a set of loosely coupled file servers to function as one. With Glamour the logical location of a file in the shared namespace is separate from the physical location at a fileserver. This simplifies the user experience by presenting a consistent filesystem view when traversing through files physically located on multiple fileservers. Furthermore, Glamour supports smart data management by replicating sets of data at custom granularities for both load balancing and failover. Glamour was made available as the Network Data Access Facility (NDAF) offering on AIX 5.3 and has been prototyped on Linux.

Glamour Namespace

As a first step towards building globally federated file services, Glamour used the NFSv4 protocol's referral mechanism to provide a common namespace across multiple file servers. This initial scope was further extended to support multiple file access protocols such NFSv3 and CIFS and multiple non-Glamour enabled platforms such as NAS filers. We have ongoing collaboration with NetApp to define possible standards for interacting with NAS filers.

In any wide-area distributed federation, performance becomes a key issue. One technique for improving performance is to support remote data caching. This enables clients to access the remote data from a local server's cache without incurring WAN latencies. Glamour's remote data cache is designed for both scalability and interoperability. For true scalability the cache data needs to be stored persistently in a clustered filesystem and accessed in parallel from the remote site while maintaining cache consistency. Ongoing work at research is exploring how best to build such a caching system.

arrow image IBM Almaden Research - File Systems

Technical Papers

Link to content in pdf format Glamour: A Wide-Area Filesystem Middleware Using NFSv4, Renu Tewari, Jonathan M. Haswell, Manoj P. Naik, Steven M. Parkes, IBM Research Report, RJ10368, July 2005.

Link to content in pdf format TAPER: Tiered Approach for Eliminating Redundancy in Replica Synchronization, Navendu Jain, Mike Dahlin, and Renu Tewari, FAST '05, December 14-15, 2005.

Nache: Design and Implementation of a Caching Proxy for NFSv4, Ajay Gulati, Manoj Naik, and Renu Tewari, FAST '07, February 13-16, 2007.

Presentations

Link to content in pdf format Glamour: An NFSv4-based File System Federation, Jonathan M. Haswell, NAS Industry Conference, October 19, 2005.


    About IBMPrivacyContact