S. R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The Global File System. In Proceedings of the 5th NASA Goddard Conference on Mass Storage Systems and Technologies, pages 319--342, College Park, MD, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....present an evaluation of the performance of the Proboscis prototype. The performance goal of Proboscis is to have a performance close to that of local disks and to have a small overhead on the nodes hosting the disks. We evaluate these goals using both raw device access and the Global file system [34] (GFS) a shared disk distributed file system with journaling) 4.1. Test Environment In the measurements, two kinds of nodes were used: fast four nodes with 933 MHz P III processors, 512 MB RAM, 66MHz 64 bit PCI bus, ServerWorks HE III chip set and two disks per node with an average seek time ....

....High Priority Client 25.3 13.4 8.0 (read) Low Priority Client 2.3 9.8 15.0 (write) Table 1. Client Bandwidth in MB s during Concurrent Access using Priority Sharing Extensions with Two Fast Clients and Single Fast Disk Node 4.3. Global File System Performance OpenGFS [30] based on GFS [34, 31]) is a serverless distributed file system, where it is assumed that all clients have direct access to all storage devices. Thus, OpenGFS is targeted at clusters using a SAN, e.g. Fibre Channel, where all nodes share the storage devices through a dedicated network. OpenGFS uses a block based ....

S. Soltis, T. M. Ruwart, and M. T. O'Keefe. The global file system. In Proc. of the Fifth NASA Goddard Conference on Mass Storage Systems and Technologies, Sept. 1996.

....independent, network aware, disk array systems such as SANs and NAS. By virtue of being server less, although these systems can provide high throughput, scalability and transparency, issues such as data sharing, security and fault management remain unaddressed. Currently, several research groups [1, 2] are working on designing intelligent storage devices capable of handling these tasks. This paper addresses the issue of maintaining coherence of data shared by multiple clients in a distributed environment. Classical reader writer synchronization allows readers shared access to common data ....

....systems, this task may be shared by clients and the storage device itself. The placement of locking mechanisms on storage devices is not a new concept. Currently, device locks (dlocks) are implemented on SCSI disks [3] and are used by distributed file systems such as the Global File System (GFS) [2]. D locks are fine grained locks that are held for relatively small periods. A GFS user must acquire a d lock in exclusive write or shared read mode before accessing storage data. However, d locks are not scalable and lead to spin locking which increases traffic on the network. Other existing work ....

S. R. Soltis, T. M. Ruwart, M. T. O'Keefe, "The Global File System," In Proc. of the Fifth NASA Goddard Space Flight Center Conference on Mass Storage Systems and Technologies, College Park, MD, September 1996.

....systems, such as NFS, cannot distribute user data across multiple nodes, some recent client server distributed file systems, such as xFS [8] Zebra [9] and Frangipani [7] use multiple nodes for improved scalability. Alternatively, a second approach is symmetric shared file systems, such as GFS [3, 4, 20]. GFS allows every node equal access to all disks directly. The third alternative, asymmetric shared file systems, supports partial disk sharing. This approach was used in HAMFS. In this approach, a dedicated node manages disk blocks containing metadata, but all other disk blocks, containing ....

....distributed file system. They conclude that their client server distributed file system (Calypso) provides much better performance than a symmetric shared file system. In our paper we have shown how an asymmetric shared file system can outperform a distributed file system organization. GFS [3,4, 20] is an example of a symmetric shared file system. It proposes a special hardware feature in the disk providing multiple logical locks. However, an asymmetric shared file system can accommodate off the shelf disk devices. Additionally, we expect GFS suffers from low performance as a result of heavy ....

S.R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The global file system. In proceedings of the 5th NASA Goddard Mass Storage Systems and Technologies Conference, College Park, MD., September 1996.

....commands but also attribute set and get are executed at the drive [G 96, RG96, G 97] The file manager is responsible primarily for verifying credentials and establishing access tokens. Soltis Global File System (GFS) uses Fibre Channel disk drives modified to support a lock primitive [SRO96, Sol97] This provides simple, efficient distributed locking. The drive itself attaches no meaning to the locks; by convention among the clients, they are used to lock inodes and other data structures. The Petal distributed disk system provides a virtual disk model on which the Frangipani ....

Steven R. Soltis, Thomas M. Ruwart, and Matthew T. O'Keefe. The global file system. In Kobler [Kob96], pages 319--342.

....media technologies, Netstation component communication is based on the TCP IP protocol suite. This design choice is controversial, in that most commercial NAP efforts (e.g. Tandem s ServerNet [15] Fibre Channel and HiPPI disks and disk arrays) and some research projects (e.g. Minnesota s GFS [21]) use special purpose protocol stacks optimized for performance on specific media. Among research systems, the most similar effort is the High Performance Storage System NAP at Lawrence Livermore [26] Digital s Petal [18] uses UDP IP over ATM as part of a large distributed virtual disk system ....

S. R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The global file system. In Kobler [17], pages 319--342.

....storage networks to support data intensive computing. Given these fast, low cost, switched networks, a serious review of the division of responsibilities between clients, servers, and storage devices has lead us to an alternative storage architecture, based upon our Global File System (GFS) [SoR96] design that is serverless and consists only of clients and networked storage devices. This proposal motivates the design of the GFS, outlines its basic structure, describes the current software and related performance results and how we intend to extend and test the GFS within the context of an ....

S. Soltis, T. Ruwart, and M. O'Keefe, "The Global File System," Fifth NASA Goddard Conference on Mass Storage Systems and Technologies, College Park, MD, September 1996.

....to post process large scientific datasets [11] on Silicon Graphics (SGI) hardware. Allowing machines to share devices over a fast Fibre Channel network required that we write our own shared file system for IRIX (SGI s System V UNIX variant) and our initial efforts yielded a prototype described in [12]. This implementation used parallel SCSI disks and SCSI reserve and release commands for synchronization. Reserve and release locked the whole device, making it impossible to support simultaneous file metadata accesses to a disk. Clearly, this was unacceptable. This bottleneck was removed in our ....

....paper) the current implementation including the details of our journaling code, new scalability results, changes to the lock specification, and our plans for GFS 5, including file system resizing. 2 GFS Background For a complete description of GFS 3 see [7] for GFS 2 see [6] and for GFS 1 see [12]. In this section we provide a summary of the key features of the Global File System. 2.1 Dlocks Device Locks are mechanisms used by GFS to synchronize client access to shared metadata. They help maintain metadata coherence when metadata is cached by several clients. The locks are implemented on ....

Steven R. Soltis, Thomas M. Ruwart, and Matthew T. O'Keefe. The Global File System. In The Fifth NASA Goddard Conference on Mass Storage Systems and Technologies, volume 2, pages 319--342, College Park, Maryland, March 1996.

.... file system [KrL86] DEC87] 1984) High Performance File Server [ArB93] 1993) Cray s Shared File System [Mat95] 1994) IBM s Parallel Journaled File System [DeM95] 1995) NASD (Network Attached Secure Disk) File Systems [Gib96] Gib97] 1995) Global File System [SoE97a] SoE97b] [SoR96] (1995) Veritas Cluster File System (CFS) 1998) Our definition of a shared file system is simple and therefore fairly broad: a shared file system allows direct data transfers between computers (clients) transferring data and the storage device that contains the data such that more than one ....

S. Soltis, T. Ruwart, and M. O'Keefe, "The Global File System," Fifth NASA Goddard Conference on Mass Storage Systems and Technologies, College Park, MD, September 1996.

No context found.

S. R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The Global File System. In Proceedings of the 5th NASA Goddard Conference on Mass Storage Systems and Technologies, pages 319--342, College Park, MD, 1996.

No context found.

S. R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The Global File System. In Proceedings of the 5th NASA Goddard Conference on Mass Storage Systems and Technologies, pages 319--342, College Park, MD, 1996.

No context found.

S. R. Soltis, T. M. Ruwart, and M. T. O'Keefe. The Global File System. In Proceedings of the 5th NASA Goddard Conference on Mass Storage Systems and Technologies, pages 319--342, College Park, MD, 1996.

No context found.

Steven R. Soltis, Thomas M. Ruwart, and Matthew T. O'Keefe. The Global File System. In Proceedings of the Fifth NASA Goddard Conference on Mass Storage Systems, pages 319--342, College Park, MD, 1996. IEEE Computer Society Press.

No context found.

Steven R. Soltis, Thomas M. Ruwart, Matthew T. O'Keefe, "The Global File System", In Proceedings of the Fifth NASA Goddard Space Flight Center Conference on Mass Storage Systems and Technologies, College Park, MD, September 1996.

No context found.

Soltis, Steven R., Ruwart, Thomas M., et al. The Global File System, proc. of the Fifth NASA Goddard Conference on Mass Storage Systems, IEEE, 1996.

No context found.

S. R. Soltis, T. M. Ruwart, M. T. O'Keefe, "The Global File System," Proceedings of the Fifth NASA Goddard Space Flight Center Conference on Mass Storage Systems and Technologies, September 17-19, 1996, College Park, MD.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC