Jehan-François Pâris

Zero-delay broadcasting protocols for video-on-demand

Broadcasting protocols for video-on-demand continuously retransmit videos that are watched simultaneously by many viewers. Nearly all broadcasting protocols assume that the client set-top box has enough storage to store between 48 and 60 minutes of video. We propose to use this storage to anticipate the customer requests and to preload, say, the first 3 minutes of the top 16 to 20 videos. This would provide instantaneous access to these videos and also eliminate the extra bandwidth required to handle compressed video signal. We present two broadcasting protocols using partial preloading to eliminate this extra bandwidth. The first of these protocols, Polyharmonic Broadcasting with Partial Preloading (PHB-PP), partitions each video into between 20 and 160 segments of equal duration and allocates a separate data stream to each individual segment. Our second protocol, the Mayan Temple Broadcasting protocol, uses fewer data streams but requires more overall bandwidth.

Evaluating the Impact of Irrecoverable Read Errors on Disk Array Reliability

We investigate the impact of irrecoverable read errors--also known as bad blocks--on the MTTDL of mirrored disks, RAID level 5 arrays and RAID level 6 arrays. Our study is based on the data collected by Bairavasundaram et al. from a population of 1.53 million disks over a period of 32 months. Our study indicates that irrecoverable read errors can reduce the mean time to data loss (MTTDL) of the three arrays by up to 99 percent, effectively canceling most of the benefits of fast disk repairs. It also shows the benefits of frequent scrubbing scans that map out bad blocks thus preventing future irrecoverable read errors. As an example, once-a-month scrubbing scans were found to improve the MTTDL of the three arrays by at least 300 percent compared to once-a-year scrubbing scans.

The performance of available copy protocols for the management of replicated data

Abstract Available copy protocols guarantee the consistency of replicated data objects against any combination of non-Byzantine failures that do not result in partial communication failures. While the original available copy protocol assumed instantaneous detection of failures and instantaneous propagation of this information, more realistic protocols that do not rely on these assumptions have been devised. Two such protocols are investigated in this paper: a naive available copy (NAC) protocol that does not maintain any state information, and an optimistic available copy (OAC) protocol that only maintains state information at write and recovery times. Markov models are used to compare the performance of these two protocols with that of the original available copy protocol. These protocols are shown to perform nearly as well as the original available copy protocol, which is shown to perform much better than quorum consensus protocols.

Using device diversity to protect data against batch-correlated disk failures

Batch-correlated failures result from the manifestation of a common defect in most, if not all, disk drives belonging to the same production batch. They are much less frequent than random disk failures but can cause catastrophic data losses even in systems that rely on mirroring or erasure codes to protect their data. We propose to reduce impact of batch-correlated failures on disk arrays by storing redundant copies of the same data on disks from different batches and, possibly, different manufacturers. The technique is especially attractive for mirrored organizations as it only requires that the two disks that hold copies of the same data never belong to the same production batch. We also show that even partial diversity can greatly increase the probability that the data stored in a RAID array will survive batch-correlated failures.

Multiple view perspectives: improving inclusiveness and video compression in mainstream classroom recordings

Multiple View Perspectives (MVP) enables deaf and hard of hearing students to view and record multiple video views of a classroom presentation using a stand-alone solution. We show that deaf and hard of hearing students prefer multiple, focused videos over a single, high-quality video and that a compacted layout of only the most important views is preferred. We also show that this approach empowers deaf and hard of hearing students by virtue of its low cost, flexibility, and ease of use in the classroom.

A stochastic approach to file access prediction

Most existing studies of file access prediction are experimental in nature and rely on trace driven simulation to predict the performance of the schemes being investigated. We present a first order Markov analysis of file access prediction, discuss its limitations and show how it can be used to estimate the performance of file access predictors, such as First Successor, Last Successor, Stable Successor and Best-k-out-of-n. We compare these analytical results with experimental measurements performed on several file traces and find out that specific workloads, and indeed individual files, can exhibit very different levels of non-stationarity. Overall, at least 60 percent of access requests appear to remain stable over at least a month.

Improving Disk Array Reliability Through Expedited Scrubbing

Disk scrubbing periodically scans the contents of a disk array to detect the presence of irrecoverable read errors and reconstitute the contents of the lost blocks using the built-in redundancy of the disk array. We address the issue of scheduling scrubbing runs in disk arrays that can tolerate two disk failures without incurring a data loss, and propose to start an urgent scrubbing run of the whole array whenever a disk failure is detected. Used alone or in combination with periodic scrubbing runs, these expedited runs can improve the mean time to data loss of disk arrays over a wide range of disk repair times. As a result, our technique eliminates the need for frequent scrubbing runs and the need to maintain spare disks and personnel on site to replace failed disks within a twenty-four hour interval.

A stream tapping protocol involving clients in the distribution of videos on demand

We present a stream tapping protocol that involves clients in the video distribution process. As in conventional stream tapping, our protocol allows new clients to tap themost recent broadcast of the video they are watching. While conventional stream tapping required the server to send to these clients the part of the video they missed, our protocol delegates this task to the clients that are already watching the video, thus greatly reducing the workload of the server. Unlike previous solutions involving clients in the video distribution process, our protocol works with clients that can only upload video data at a fraction of the video consumption rate and includes a mechanism to control its network bandwidth consumption.

Self-adaptive disk arrays

We present a disk array organization that adapts itself to successive disk failures. When all disks are operational, all data are mirrored on two disks. Whenever a disk fails, the array reorganizes itself, by selecting a disk containing redundant data and replacing these data by their exclusive or (XOR) with the other copy of the data contained on the disk that failed. This will protect the array against any single disk failure until the failed disk gets replaced and the array can revert to its original condition. Hence data will remain protected against the successive failures of up to one half of the original number of disks, provided that no critical disk failure happens while the array is reorganizing itself. As a result, our scheme achieves the same access times as a mirrored organization under normal operational conditions while having a much lower likelihood of loosing data under abnormal conditions. In addition it tolerates much longer repair times than mirrored disk arrays.

Using shared parity disks to improve the reliability of RAID arrays

We propose to increase the reliability of RAID level 5 arrays used for storing archival data. First, we identify groups of two or three identical RAID arrays. Second, we add to each group a shared parity disk containing the diagonal parities of their arrays. We show that the new organization can tolerate all double disk failures and between 75 and 89 percent of triple disk failures without incurring any data loss. As a result, the additional parity disk increases the mean time to data loss of the arrays in the group it protects by at least 14,000 percent.