Dan Feng

Scalable and Adaptive Metadata Management in Ultra Large-Scale File Systems

This paper presents a scalable and adaptive decentralized metadata lookup scheme for ultra large-scale file systems (ges Petabytes or even Exabytes). Our scheme logically organizes metadata servers (MDS) into a multi-layered query hierarchy and exploits grouped bloom filters to efficiently route metadata requests to desired MDS through the hierarchy. This metadata lookup scheme can be executed at the network or memory speed, without being bounded by the performance of slow disks. Our scheme is evaluated through extensive trace-driven simulations and prototype implementation in Linux. Experimental results show that this scheme can significantly improve metadata management scalability and query efficiency in ultra large-scale storage systems.

A Covert Communication Model Based on Least Significant Bits Steganography in Voice over IP

Steganography, as one of alternative techniques for secure communications, has drawn more and more attentions. This paper presents a covert communication model based on least significant bits (LSB) steganography in Voice over IP (VoIP). The model aims at providing nice security of secret messages and real-time performance that is vital for VoIP. Therefore, we employ a simple encryption of secret messages before embedding them. This encryption strikes a good balance between adequate short-term protection for secret messages and low latency for VoIP. Furthermore, we design a structure of embedded messages. It can provide flexible length and avoid effectually both extraction attack and deceptive attack. We evaluate the model with ITU-T G.729a as the codec of the cover speech in StegTalk, our platform for study on covert communications theory in VoIP. In this case, the proposed model can provide two optional covert transmission speeds, i.e. 0.8 kb/s and 2.6 kb/s, where the maximum payload ratio is 99.98%. The experimental results show that our method has negligible effects on speech quality and well meets the real-time requirement of VoIP.

Design, implementation and performance evaluation of a cost-effective, fault-tolerant parallel virtual file system

Fault tolerance is one of the most important issues for parallel file systems. This paper presents the design, implementation and performance evaluation of a cost-effective, fault-tolerant parallel virtual file system (CEFT-PVFS) that provides parallel I/O service without requiring any additional hardware by utilizing existing commodity disks on cluster nodes and incorporates fault tolerance in the form of disk mirroring. While mirroring is a straightforward idea, we have implemented this open source system and conducted extensive experiments to evaluate the feasibility, efficiency and scalability of this fault tolerant approach on one of the current largest clusters, where the issues of data consistency and recovery are also investigated. Four mirroring protocols are proposed, reflecting whether the fault-tolerant operations are client driven or server driven; synchronous or asynchronous. Their relative merits are assessed by comparing their write performances, measured in the real systems, and their reliability and availability measures, obtained through analytical modeling. The results indicate that, in cluster environments, mirroring can improve the reliability by a factor of over 40 (4000%) while sacrificing the peak write performance by 33--58% when both systems are of identical sizes (i.e., counting the 50% mirroring disks in the mirrored system). In addition, protocols with higher peak write performance are less reliable than those with lower peak write performance, with the latter achieving a higher reliability and availability at the expense of some write bandwidth. A hybrid protocol is proposed to optimize this tradeoff.

Improved read performance in a cost-effective, fault-tolerant parallel virtual file system (CEFT-PVFS)

Due to the ever-widening performance gap between processors and disks, I/O operations tend to become the major performance bottleneck of data-intensive applications on modern clusters. If all the existing disks on the nodes of a cluster are connected together to establish high performance parallel storage systems, the clusters overall performance can be boosted at no additional cost. CEFT-PVFS (a RAID 10 style parallel file system that extends the original PVFS), as one such system, divides the cluster nodes into two groups, stripes the data across one group in a round-robin fashion, and then duplicates the same data to the other group to provide storage service of high performance and high reliability. Previous research has shown that the system reliability is improved by a factor of more than 40 with mirroring while maintaining a comparable write performance. This paper presents another benefit of CEFT-PVFS in which the aggregate peak read performance can be improved by as much as 100% over that of the original PVFS by exploiting the increased parallelism. Additionally, when the data servers, which typically are also computational nodes in a cluster environment, are loaded in an unbalanced way by applications running in the cluster, the read performance of PVFS will be degraded significantly. On the contrary, in the CEFT-PVFS, a heavily loaded data server can be skipped and all the desired data is read from its mirroring node. Thus the performance will not be affected unless both the server node and its mirroring node are heavily loaded.

Implementation and Performance Evaluation of an Object-Based Storage Device

Object-based storage system (OBSS) has led to a new wave in network storage area . Object-based storage device (OSD) which is the cornerstone of OBSS plays a decisive role in the performance of the whole OBSS. This paper describes how we implemented an object-based storage device which adopts a unique hardware architecture based on switching fabric supporting parallel data transfer in multiple I/O channels, and a new object-based storage device file system (HustOSDFS) that can reduce the software overloads to achieve high performance. The experimental evaluation results show that the OSD performs well for system performance. Furthermore, the OSD provides characteristic of large capacity and low cost.

RAID10L: A high performance RAID10 storage architecture based on logging technique

RAID10 storage system suffers the relatively poor write performance due to the write request must be served by both disks in a mirror set. To address this problem, in this paper we propose a novel RAID10 storage architecture, called RAID10L, which extends the data mirroring redundancy of RAID10 by incorporating a dedicated log disk. The goal of RAID10L is to significantly improve the write performance of RAID10 at some little expense of reliability. In RAID10L both read and write requests are processed in a balance scheme. For every write request, RAID10L keeps two copies of the write data: one in its normal place of data disk chosen by a write balance scheme and the other in the log disk by writing sequentially. The update to another data disk in a mirror set is delayed to the next quiet period between bursts of client activity. Reliability analysis shows that the reliability of RAID10L, in terms of MTTDL (mean time to data loss), is somewhat worse than RAID10 but much better than RAID5. On the other hand, our prototype implementation of RAID10L driven by Iometer benchmark shows that RAID10L outperforms RAID10 by up to 47.1% and RAID0 by 27.3% in terms of average response time. Driven by some real-life traces, RAID10L gains improvement up to 30.7% with an average of 27.7% than RAID10 in terms of average response time.

An efficient, low-cost inconsistency detection framework for data and service sharing in an Internet-scale system

In this paper, we argue that a broad range of Internet-scale distributed applications can benefit from an underlying low-cost consistency detection framework that is an alternative to inconsistency avoidance and can detect inconsistency among nodes sharing data or services in a timely manner. This paper first presents an overview of the inconsistency detection framework. Then, it discusses the detailed design of the two-layer inconsistency detection module, the core component of this framework, which can detect inconsistency among nodes in a timely manner. The proposed two-layer inconsistency detection module is evaluated both analytically and via simulations, which shows that this module can significantly reduce the time to detect inconsistency among nodes without adding much maintenance cost. Finally, this paper outlines the possible mechanisms to discern the application semantics and to resolve the detected inconsistencies

Exploiting redundancy to boost performance in a RAID-10 style cluster-based file system

While aggregating the throughput of existing disks on cluster nodes is a cost-effective approach to alleviate the I/O bottleneck in cluster computing, this approach suffers from potential performance degradations due to contentions for shared resources on the same node between storage data processing and user task computation. This paper proposes to judiciously utilize the storage redundancy in the form of mirroring existed in a RAID-10 style file system to alleviate this performance degradation. More specifically, a heuristic scheduling algorithm is developed, motivated from the observations of a simple cluster configuration, to spatially schedule write operations on the nodes with less load among each mirroring pair. The duplication of modified data to the mirroring nodes is performed asynchronously in the background. The read performance is improved by two techniques: doubling the degree of parallelism and hot-spot skipping. A synthetic benchmark is used to evaluate these algorithms in a real cluster environment and the proposed algorithms are shown to be very effective in performance enhancement.

A Novel Cost-Effective Disk Scrubbing Scheme

a distinct benefit of disk scanning or scrubbing operation is identifying the potential failure sectors as early as possible, thus providing high reliability. Obviously, the higher the scrubbing frequency is, the higher the system reliability is. However, it may take a few hours for a scanning process to check the whole disk. In other words, the scrubbing process may result in a downtime or a lower system performance. Furthermore, the scrubbing process consumes energy. In order to reduce the impact of disk scrubbing on disk performance and energy consumption, system designers choose to scan the disk in a low frequency, which results in a lower reliability. Additionally, conventional disk scrubbing schemes assume that the disk failure rate is constant, while the recent researches[2][6] show the disk failure rate is more complex. In this paper, we present a novel scrubbing scheme to solve the above challenges. In the scheme, an optimum scrubbing cycle is decided by keeping a balance between data loss cost, scrubbing cost, and disk failure rate. Our research shows that the scrubbing scheme is applicable for storage with low-capacity disk and inexpensive data.

FairOM: enforcing proportional contributions among peers in internet-scale distributed systems

The viability of overlay multicasting has been established by previous research. However, in order to apply overlay multicast to Internet-scale distributed systems, such as the Grid and Peer-to-Peer systems, the issue of effectively enforcing fairness among peers so as to optimize overall performance remains as a challenge. This paper argues that simply applying a multiple-tree scheme does not provide sufficient fairness, in terms of performance. Instead, we believe that a better way to define fairness, for performance’s sake, is to factor in peers’ proportional contributions as it provides the opportunity to support many simultaneous multicasting sessions. This paper then presents a protocol, called FairOM (Fair Overlay Multicast), to enforce proportional contributions among peers in Internet-scale distributed systems. By exploiting the notion of staged spare capacity group and deploying a two-phase multicast forest construction process, FairOM enforces proportional contributions among peers, which enables more simultaneous multicasting sessions and alleviates potential hot-spots. The simulation results of a large multicast group with 1000 members show that FairOM achieves the goal of enforcing proportional contributions among peers and does not overwhelm the peers, including the multicast source. FairOM also achieves low delay penalty for peers and high path diversity.