NAS Conference Talks 2003


Home Announcements Documents Events Links Talks

Quick Jump to NAS Conference Talks

2005 2004 2003 2002 2001 2000
Why NFSv4 ? Spencer Shepler, Sun Microsystems
This tutorial will cover the feature content and concept of NFS version 4. Topics covered include justification for deployment of NFSv4, with respect to delegation, security, and continual evolution of the protocol. Special considerations for deployment within Fortune 500 companies will be addressed.
NFS in the Microsoft World: An Introduction to NFS Integration Robert Wong, Hummingbird
As a foreign protocol to the Microsoft architecture, how can NFS be able to efficiently integrate to the operating system, while remaining transparent to the end users? This tutorial will look into various aspects of the integration, and how NFSv4 provides a better fit with the Windows file system model.
AFS/DFS to NFSv4 Migration Issues Robert Thurlow, Sun Microsystems
This tutorial will cover the major issues when migrating from an AFS to and NFSv4 environment. Automounting, caching, security, replication and OpenAFS will be discussed, along with coming changes in NFSv4 that will make the transition more seamless.
Addressing Multiprotocol File Sharing Environments Jeff Purcell, EMC
A common challenge faced by customers today is handling filesharing within environments that are predominantly multi-protocol (heterogeneous environments). Traditionally file sharing environments provide access to files in a sharing capacity via NFS, CIFS or other filesystem protocol. This tutorial will expose the benefits of leveraging NAS in environments requiring access to the same data shared via multiple protocols (NFS and CIFS). Specifically, how to present the same data to network clients via NFS, CIFS and FTP, a variety of issues associated with maintaining proper security (including locking symantics) and managing the interaction between a UNIX and a Windows environment when deploying NAS in a multi-protocol environment will be discussed.
NFS over RDMA Sweet Spot Brent Callaghan, Sun Microsystems
RDMA is a new technology that will accelerate NFS data over high speed networks. This tutorial will describe how NFS over RDMA works, then discuss some scenarios that will make most effective use of this new technology. Considerations such as application types, access patterns, kinds of hardware and networks and administration will be presented.
Database over NFS Brian Pawlowski, Network Appliance
Considerations for setup and deployment of databases using NFS. Covering architecture, performance tuning, availability, backup, disaster recovery, scalability. Answering the questions: Why NAS? Why DB over NAS? What benefits are derived? Common problems described.
NFS Performance Concerns Peter Staubach, Sun Microsystems
This tutorial will cover general NFS performance issues from a platform independent view. Topics will include some of the system level and networking issues that can affect the performance of applications accessing files on NFS mounted file systems. Some of the common tuning parameters for NFS access will be examined for their effects.
Sun Microsystems Keynote: From NFS to NFS Rob Gingell, Sun Microsystems Fellow and Chief Engineer
Shared File System Use in ETL and Data Warehousing Dan Pollack, AOL
Many companies have large data warehouse environments that require the use of many incoming data sources. These data sources are often combined together with extraction, transformation, and load (ETL) tools and then loaded into multiple databases for analysis. There is significant advantage to storing these processed data sets in a central location for use by multiple databases and for later retrieval in the event of a data quality problem such as corruption or schema change. This environment tends to be strenuous for NFS based systems due to a number of factors. These factors include high throughput performance requirements, large file sizes, and high numbers of shared file systems in a heterogeneous client and server environment.
I will discuss AOL's experience with these issues, how we deal with them, and what improvements we feel will help most with our shared storage environment.
NFSv4 Open Source Project Update Andy Adamson, University of Michigan
Andy will present an update on the Linux and BSD NFSv4 implementations including a road-map for NFSv4 inclusion into the Linux and BSD kernels.
NFS Client Benchmarking John Corbin, EP Network Storage Performance Lab
Several NFS server vendors have continually optimized their implementation over the past decade to the point where NFS client implementations now appear to be the performance bottleneck. While application issues may dominate an end users choice of which NFS client to use, the vendors of NFS clients should be interested in making sure they have the best performance possible. This presentation will look at the issues that exist in NFS client benchmarking, the different approaches that could be taken, and a prototype NFS client benchmark.
The Data Management Challenge Steven Kleiman, Network Appliance Senior Vice President and CTO
NFS has come a long way from its roots as a way of sharing files in home directories over 10Mbps unswitched networks. Enterprise customers now use it to access large databases, Petabyte datasets, and for shared storage in volume server farms using 1Gbps switched networks. Such networks bring new challenges of scale, reliability and manageability. The NFS protocol is evolving to meet the challenge of managing this challenge.
Global Storage Architecture: Scalable NFS Service Using off the Shelf Components Stanley Wood, IBM
The Global Storage Architecture (GSA) is a system that provides scalable and robust file system service to commodity clients. It uses off the shelf components including IBMs GPFS cluster file system to provide file service to standard Windows and NFS clients. The clustered back end provides many of the benefits of file systems like AFS and DFS, like location independence, improved scalability, and improved reliability without using proprietary client software. GSA is in active use inside IBM with servers deployed at 11 sites worldwide and counting.
Sun N1: Storage Virtualization and Oracle Glenn Colaco/Richard McDougall, Sun Microsystems
As customers RDBMSs continue to grow, their storage subsystem complexity also continues to grow. New technologies are created to address these issues, however data manageability is still the number one issue customers continue to face today. Storage virtualization via file based Network Attached Storage protocols have been addressing this issue head on. While utilizing existing infrastructure, tried and true tools/diagnostic facilities, and additional feature sets such as end-to-end data security and Quality of Service, NAS is becoming critical to databases.
NAS has had a long successful history, but the number one issue precluding NAS from being widely accepted is the perceived performance problems on RDBMSs. The Sun Oracle-NAS project is addressing these performance problems, allowing customers to successfully use NAS with Oracle. By evaluating other protocols and transports including DAFS and RDMA, as well as next generation transports such as Infiniband, the Sun Oracle-NAS project aims to make NAS a viable alternative to block based/SAN storage.
As part of Suns N1 strategy, NAS protocols become a critical utility needed for network based computing. It is especially important when dealing with blade servers, and virtualized compute resources where commercial applications can run on any node.
Rehabilitating NFS Security Michael Eisler, Network Appliance
NFS has been known for bad security. This talk examines the basis for that perception, recounts the excellent progress to improve NFSs image, and suggests how to prevent a relapse in the ongoing rehabilitation of NFS security.
Benefits of full TCP/IP offload (TOE) for NFS Services Hari Ghadia, Adaptec
While TCP/IP offload engine (TOE) products will accelerate a range of network applications, NFS is emerging as a sweet-spot TOE application due to its network traffic profile. The presentation will discuss how TOEs can help scale file serving performance, reduce client access time, and increase in number of clients served by a given file server. Customer case study deploying TOE NICs in clustered environment will be discussed.
EMC Keynote: Data Protection in an Enterprise Environment John Hayden, EMC Corporation Prinicpal Software Engineer
NFS Version 4 Spencer Shepler, Sun Microsystems
The April release of RFC3530 sets the NFSv4 protocol as a IETF Proposed Standard. Now that the NFSv4 protocol definition is complete and implementations are being released, what is next for the IETF NFSv4 working group. Spencer will review the NFSv4 working group charter and recent activities surrounding the use of the NFSv4 minor versioning mechanism.
The Death of CIFS Jeremy Allison, Samba Team
Microsoft is talking about moving towards a new remote file system protocol in their "Longhorn" Windows release. What might this look like? What effect will this have on the CIFS community? What (if any) alternatives do we have to keep creating value in the Network Attached Storage space?
A Roadmap for NFS on RDMA Thomas Talpey, Network Appliance
Recently, several proposals have been put forward for NFS, especially for NFS version 4, to employ Remote Direct Memory Access (RDMA) transport technology. This will bring NFS performance to an entirely new level, at the same time it enables further enhancements, such as direct application use of NFS, enhanced local sharing semantics, and improved operation even for non-RDMA enabled peers.
This talk will explore a roadmap for NFS enhancements going forward with RDMA technologies. The distinct, yet related benefits from each aspect of the proposals will be discussed. Timeframes for the availability of the various components of the solution will be outlined.
Preparing Intel for NFSv4 Deployment Travis Broughton, Intel
Over the past year, Intel has been preparing its environment for the introduction of NFSv4. We have deployed some test NFSv4 clients and fileservers in a lab setting, and have begun experimenting with its features. Our goal is to provide a corporate-wide distributed filesystem, much like AFS or DCE/DFS. Along with the NFSv4 work, we have been investigating complementary solutions such as building a namespace with autofs or am-utils, providing an authentication framework with LDAP and/or Kerberos, and various other components not supplied directly by the filesystem. This talk will outline our work to date, the challenges we've faced, and our vision of what our environment might look like a year from now.
NFS for The Enterprise Fred Whiteside, Hummingbird Senior Vice President, Research and Development
The SNIA NAS Management TWG: Its Work and Status Dennis Chapman, Network Appliance
This talk will describe the work being done by the SNIA NAS Management Technical Work Group to develop a standard way to manage NAS devices. We will cover the scoping, modeling, recipe and draft phases of this effort. The goal of this is work is to include management of NAS devices in the SNIA SMI-S.
Mixed Protocols: And the Winner is... John Terpstra, PrimaStasys, Inc
The presentation looks at the current desktop and server market, considers the profile of network file system protocols, looks at market trends, and draws some conclusions. It then considers the future of NFS and other file system protocols (i.e., CIFS) with a keen eye on changes that are taking place in how information is accessed and used across disparate platforms. The talk aims to provide a foundation for discussion to the following questions: What will the end-game in the protocol debate look like? What is the role of NFS standardization in this future landscape? IS there a consumer market for future products and services?

Questions or comments regarding this page? Tell Tom Haynes.