| High Performance and Distributed NAS Server Architecture for Scalable and Global NFS file systems | Mike Kazar, CTO, Spinnaker Networks |
| Many different architectures exist in the marketplace today to provide NFS file serving, from NAS to general-purpose systems with DAS. This presentation will discuss a new NAS architecture from Spinnaker Networks, based upon a distribute file system which enables the IT administrator to build large, high performance NFS file systems with a global name space. | |
| NFS version 4 | Spencer Shepler, Sun Microsystems |
| RFC3010 was the start. Implementations were started, issues were found. The update to RFC3010 (NFS version 4) is complete and coming to an RFC list near you. Will take a look at where the NFS version 4 implementation community is currently at and what changed from RFC3010 and where the NFS version 4 standard is going. | |
| NFSv4 Open Source Reference Implementation Project Update | Andy Adamson, CITI-University of Michigan |
| Andy will present an update on both the Linux and OpenBSD NFSv4 implementations including namespace construction, identity resolution, ACL mapping, and a road-map for NFSv4 inclusion into the Linux and OpenBSD kernels. | |
| Hummingbird Keynote: Hummingbird NFS Maestro = NFSv4 | Dan Trufasiu, Director of Connectivity R&D, Hummingbird |
| What does the NFS protocol offer the Windows PC world? What is the bottom line on NFSv3? Topics discussed will include Hummingbird's implementation of NFSv4: what has been done, how it was done, and the restrictions of the implementation. | |
| NFS at 10Gb/sec | Brent Callaghan, Sun Microsystems |
| While NFS enjoys the ubiquity and interoperability of TCP/IP networking, it is facing a growing performance barrier inherent in the architecture of host networking stacks. Clients and servers are unable to keep up with a "fire hose" of data. RDMA provides direct data placement and network offload into hardware, decoupling network performance from host performance. NFS can use RDMA to take full advantage of the performance benefits of multi-Gigabit Ethernet and Infiniband networks. This presentation will describe RDMA, how NFS will use it, and how it will affect the balance of SAN vs NAS storage. | |
| Network Adapters for NFS at 10Gb and Beyond | Tom Hotchkiss VP, Engineering, Emulex |
| NFS will require Network Adapter support to run at 10Gb and beyond. Hardware offload and new protocols such as RDMA will be required to address the shortcomings of today's approach. Multi-protocol Network Adapters interoperate with existing protocols and provide a migration path to RDMA and other future technologies. | |
| NFS Clustered Filesystem | Xiaoye Jiang, EMC |
| Scaling NFS server throughput can be accomplished in several ways; most commonly NFS servers approach the challenge of scaling by devoting larger and larger machines to the task, which creates larger single points of failure. An alternate approach to achieving scalability is using multiple servers in a cluster configuration. This approach has not been widely adopted because of the challenge of maintaining cache coherency between the nodes. Leveraging MPFS technology natively on EMC Celerra NFS servers allows you to build incrementally scalable amounts of front-end capacity to a shared filesystem. This allows you to treat your NFS serving requirements much like you treat your storage requirements, scaling incrementally to meet your front-end I/O requirements. | |
| NFS v4 - Customer problems and NetApp's solutions | Amit Sinha, Multiprotocol Product Marketing Manager, Network Appliance |
|
A brief overview of typical customer NFS environment, their current
challenges with NFS v2 and v3. It will then lead to how NFS v4 will
benefit these customers and their environments, for example security,
performance.
In addition, the presentation will also focus on how NetApp is driving the NFS v4 development efforts with partnerships from leading NFS v4 client vendors. | |
| Migrating from AFS to NFS for Global Data Sharing | Travis Broughton, Intel |
| For the past decade, Intel has been using AFS to share design data globally, with NFS (protocol versions 2 and 3) used to provide higher performance for data with no inter-site access requirements. With the EOL of AFS announced by IBM, we have been examining other technologies to provide global access to data, including OpenAFS, NFS (protocol versions 3 and 4), and caching NFS proxy solutions. This presentation will describe the existing AFS-based environment, the advantages we see in NFS (especially NFSv4), and the challenges -- both technical and political -- that we are facing when considering migrating to NFS-based solutions. | |
| NFSv4 Vendor's Discussion: Notes | The Vendors |
| EMC Keynote: Policy Based Management and Tiered Backends | John Hayden, EMC |
| Storage platforms are becoming more and more a commodity available at specific tiered levels - running the gamut from JBOD to simple RAID arrays, to complete storage subsystems with integrated features such as replication and snapshots. NAS devices, attached to these different tiers of storage from many different vendors, are presented with new and unique management and functionality challenges. Tools and strategies to manage the transparent storage provisioning and migration of data from one tier to another, as well as policy based engines to automatically match business requirements (service level agreements, recoverability, replication, performance, availability etc.) to the available storage tiers are needed in such environments. | |
| Analyzing NFS Client Performance with IOzone | Tom McNeal, TRM Consultants and Don Capps, Hewlett-Packard |
| This paper will review the basic structure of IOzone, a file system benchmark which may be used for assessing client performance issues in an NFS client/server environment. It will review IOzone's use of file IO for its base measurements, and will discuss possible system and network variations based on IO type, SMP, cache management, process vs. thread usage, transport type, etc. It will also review the system parameters useful in modifying system behavior when testing with IOzone, in order to investigate client behavior in the most extensive and illustrative way. | |
| NFS: Pushing the Limit | Smita Thakur, Sun Microsystems |
| Deploying Scalable Switched NAS Infrastructures within NFS Environments | Vladimir Miloushev, CTO and Founder, Z-force |
| "file switch" will enable the first "NAS Array", an unlimited number of commodity, low-cost NAS boxes that can be connected together to provide aggregated performance and unlimited scalability at up to wire speeds, in an environment that is truly manageable as a single entity. This session will discuss the file switch in more detail and the challenges with switching file transactions at wire speeds. It will look at the NFS protocol and its strengths and limitations for large fileserving environments. The session will look at the unique issues associated with NFS interoperability, security and locking mechanisms to provide an aggregated NAS environment. | |
| Network Appliance Keynote: Storage Networking and NFS | Brian Pawlowski, Vice President, Network Appliance |
| NFS is one of several options a customer has for storage networking today. What are its strengths? What applications are best suited to NFS? Where does NFS face challenges? This talk will put NFS in the broader context of storage networking. | |
| Managing NFS in Solaris using CIM/WBEM | Sarah Jelinek, Sun Microsystems |
| CIM/WBEM is a technology that is gaining ground in the data resource management area. Many vendors, including Sun Microsystems, are planning on rolling out data resource management products based on the DMTF CIM and SNIA Bluefin specifications. NFS in Solaris will be fully manageable via CIM/WBEM. This talk will present the full Solaris NFS CIM model, along with examples on how to use it in Solaris. | |
| Replication and Migration | Rob Thurlow, Sun Microsystems |
| Replication and migration of data is critical functionality for many. As NFS Version 4 nears standardization, the IETF NFSv4 working group focus is shifting to this area with a new replication/migration protocol as well as changes to the core protocol to better support it. In this presentation, you will hear about the recent progress and current status of this work. | |
| DAFS Extensions for NFS | Mark Wittle, Network Appliance |
| DAFS, the Direct Access File System, extends NFS version 4 with additional capabilities. The presentation will focus on the motivation, approach, and implementation of these features. | |