J.R. Douceur, R.P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System," In IEEE SRDS, 2001  Home/Search   Document Not in Database   Summary   Related Articles   Check

....per block written. Pangaea [20] allows individual servers to continue serving most of their data even when disconnected; however, this approach is designed for widearea systems, and does not permit files to be striped across dozens of servers for higher bandwidth. Microsoft s Farsite project [1, 5] investigated the issue of replication for systems built from relatively unreliable commodity workstations. They focused on reliability, investigating replica placement strategies that take server reliability into account. However, Farsite is not designed for high bandwidth applications, and must ....

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of the 20th Symposium on Reliable Distributed Systems (SRDS '01), pages 4--13, New Orleans, LA, Oct. 2001. IEEE.

....by IBM Faculty Award. Linux implementation of several principals of CODA using existing (available) infrastructure. In addition, the peer to peer file system OceanStore [1] provides a secure (using encryption) and durable global storage over the Internet, allowing a public storage model. Farsite [5] is a scalable, server less file system that exploits storage and communication resources of desktop computers within a large organization, dynamically managing replication by monitoring the organization computers. Our contribution: We propose the first self stabilizing distributed file system. ....

J. Douceur and R. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In In the Proceedings of the 20th Symposium on Reliable Distributed Systems (SRDS), New Orleans, Louisiana, October 2001. http://distcomp.ethz.ch/publications/SRDS01.pdf.

....designed to serve as a replacement for a traditional Web cache. While such detailed protocol design and implementation issues are clearly important, the work [5] has not focused on the critical issues of replication and object replacement policies for a P2P cache. FarSite [15] see also [16] [17]) is a P2P file system with the strong persistence and availability of a traditional file system. The FarSite filesystem uses the same number of replicas for each object. In contrast with a file system, the goal of a P2P cache is not to provide strong file persistence and availability, but instead ....

J.R. Douceur, R.P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System," Proceedings of 20th IEEE SRDS, 2001

....users. That is, the server separately maintains file updates for the users. The forked users never see each other s changes, and each user believes its state reflects reality. A higher level Byzantine agreement protocol, which is potentially expensive, might be necessary to address this issue [11]. Recently Mazieres and Shasha [32] introduced the notion of fork consistency and a protocol to achieve it. Though their scheme does not prevent a forking attack, it makes it easier to detect. 5 Protocols We now summarize the steps involved in protocols for creating, reading and writing as well ....

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In SRDS, pages 4--13, 2001.

....of the classic model may lead to inaccurate estimations. The definition of independence is, P (Node X down I States of other nodes in the system) P (Node X down) The independence assumption is not always valid. Several studies show that storage node failures are correlated. In [Bolosky2000][Douceur2001c], the actual scheme availabilities (replication in their case) differ from the calculations that assume independent failures. In [Amir1996] they analyze 14 machines in a wide area network and report that crashes are correlated. Our measurements of web server availabilities also support the fact ....

....of machines. They only model 2 way correlations, whereas in our case, to determine the availabilities of schemes with n 2, in addition to 2 way correlations higher way correlations are also modeled. Therefore, their model does not involve or require any estimation of unknown probabilities. In [Douceur2001c], effective system availability (ESA) is used to compare the availabilities of different schemes. The availabilities of individual files are calculated using the independence assumption. ESA is a weighted average of the availabilities of the files in the system. Files with low availabilities have ....

[Article contains additional citation context not shown here]

J. R. Douceur, R. P. Wattenhofer "Optimizing File Availability in a Secure Serverless Distributed File System", Proceedings of20th IEEE SRDS, pp. 4-13, 2001

....is fundamentally untrusted. For example, CU employs a trust model with different levels of trust to grant privileges to other machines. This is superior to systems such as OceanStore that use only cryptographic techniques to protect the data. 4. 3 Farsite The Farsite project at Microsoft research [23] is a serverless, distributed file system that exploits the underutilized storage and communication resources distributed among the networked desktop computers in a large organization. Its distinguishing characteristic is that it can be deployed on an existing desktop infrastructure and does not ....

John R. Douceur and Roger P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System", Proceedings of the 20 IEEE Symposium on Reliable Distributed Systems, New Orleans, LA (October 2001), pp. 4-13.

No context found.

J. R. Douceur and R. P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System", 20th SRDS, IEEE, Oct 2001, pp. 4-13.

....and controlled replication causes a dramatic increase in the space consumed by the file system, it is advantageous to reclaim storage space due to incidental and erratic duplication of file content. Since the disk space of desktop computers is mostly unused [13] and becoming less used over time [14], reclaiming disk space might not seem to be an important issue. However, Farsite (like most peer to peer systems [32] relies on voluntary participation of the client machines owners, who may be reluctant to let their machines participate in a distributed file system that substantially reduces ....

....3) Relocating the replicas of files with identical content to a common set of storage machines. 4) Coalescing the identical files to reclaim storage space, while maintaining the semantics of separate files. The latter two of these problems are addressed by a file replica relocation system [14] and the Windows 2000 [34] Single Instance Store (SIS) 7] which are described in other publications. In the present paper, we describe the first two problems solutions: 1) convergent encryption, a cryptosystem that produces identical ciphertext files from identical plaintext files, irrespective ....

[Article contains additional citation context not shown here]

J. R. Douceur and R. P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System", 20th SRDS, IEEE, Oct 2001, pp. 4-13.

....operational. The increase in the (light) metadam workload on high availability machines is compensated by a small decrease in their (heavy) file storage and replication workload. Farsite improves global file availability by continuously relocating file replicas at a sustainable background rate [14]. Overall mean file uptime is maximized by achieving an equitable distribution of file availability, because low availability files degrade mean file uptime more than high availability files improve it. Therefore, Farsite successively swaps the machine locations of replicas of high availability ....

....files degrade mean file uptime more than high availability files improve it. Therefore, Farsite successively swaps the machine locations of replicas of high availability files with those of lowavailability files, which progressively equalizes file availability. Simulation experiments [14] driven by actual measurements of desktop machine availability show that Farsite needs to swap 1 of file replicas per day to compensate for changes in machine availability. File availability is further improved by caching file data on client disks. These caches are not fixed in size but rather ....

[Article contains additional citation context not shown here]

J. R. Douceur and R. P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System", 20th SRDS, IEEE, Oct 2001.

No context found.

J.R. Douceur, R.P. Wattenhofer, "Optimizing File Availability in a Secure Serverless Distributed File System," In IEEE SRDS, 2001

No context found.

J. Douceur and R. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of the 20th IEEE SRDS, pages 4--13, New Orleans, LA, October 2001.

No context found.

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proc. of SRDS, 2001.

No context found.

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of the 20th Symposium on Reliable Distributed Systems (SRDS '01), pages 4--13, New Orleans, LA, Oct. 2001. IEEE.

No context found.

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of the 20th Symposium on Reliable Distributed Systems (SRDS '01), pages 4--13, New Orleans, LA, Oct. 2001. IEEE.

No context found.

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of the 20th Symposium on Reliable Distributed Systems (SRDS '01), pages 4--13, New Orleans, LA, Oct. 2001. IEEE.

No context found.

J. R. Douceur and R. P. Wattenhofer. "Optimizing File Availability in a Secure Serverless Distributed File System", Proc. of 20th IEEE SRDS, 2001.

No context found.

J. Douceur and R. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proceedings of 20th IEEE SRDS, pages 4--13, 2001.

No context found.

J. Douceur and R. Wattenhofer, "Optimizing file availability in a secure serverless distributed file system," in Proceedings of 20th IEEE SRDS, 2001, pp. 4--13.

No context found.

J. R. Douceur and R. P. Wattenhofer. Optimizing file availability in a secure serverless distributed file system. In Proc. of SRDS, 2001.

No context found.

J. R. Douceur and R. P. Wattenhofer. "Optimizing File Availability in a Secure Serverless Distributed File System", Proc. of 20th IEEE SRDS, 2001.

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