Exploiting Disk Layout and Block Access History for I/O Prefetch

2010 ◽  
pp. 201-217 ◽  
Author(s):  
Feng Chen ◽  
Xiaoning Ding ◽  
Song Jiang
Keyword(s):  
Critical Role ◽  
Hard Disk ◽  
Storage Device ◽  
Access Pattern ◽  
Data Layout ◽  
Modern Computer ◽  
Access Cost ◽  
Random Block ◽  
Disk Layout ◽  
Access Patterns

As the major secondary storage device, the hard disk plays a critical role in modern computer system. In order to improve disk performance, most operating systems conduct data prefetch policies by tracking I/O access pattern, mostly at the level of file abstractions. Though such a solution is useful to exploit application-level access patterns, file-level prefetching has many constraints that limit the capability of fully exploiting disk performance. The reasons are twofold. First, certain prefetch opportunities can only be detected by knowing the data layout on the hard disk, such as metadata blocks. Second, due to the non-uniform access cost on the hard disk, the penalty of mis-prefetching a random block is much more costly than mis-prefetching a sequential block. In order to address the intrinsic limitations of filelevel prefetching, we propose to prefetch data blocks directly at the disk level in a portable way. Our proposed scheme, called DiskSeen, is designed to supplement file-level prefetching. DiskSeen observes the workload access pattern by tracking the locations and access times of disk blocks. Based on analysis of the temporal and spatial relationships of disk data blocks, DiskSeen can significantly increase the sequentiality of disk accesses and improve disk performance in turn. We implemented the DiskSeen scheme in the Linux 2.6 kernel and we show that it can significantly improve the effectiveness of filelevel prefetching and reduce execution times by 20-53% for various types of applications, including grep, CVS, and TPC-H.

Application of both Temporal and Spatial Localities in the Management of Kernel Buffer Cache

2010 ◽  
pp. 107-117
Author(s):  
Song Jiang
Keyword(s):  
Buffer Management ◽  
Direct Access ◽  
Storage Device ◽  
Data Layout ◽  
Data Intensive ◽  
Spatial Locality ◽  
Buffer Cache ◽  
Temporal Locality ◽  
Upper Level ◽  
Access Patterns

As the hard disk remains as the mainstream on-line storage device, it continues to be the performance bottleneck of data-intensive applications. One of existing most effective solutions to ameliorate the bottleneck is to use the buffer cache in the OS kernel to achieve two objectives: reduction of direct access of on-disk data and improvement of disk performance. These two objectives can be achieved by applying both temporal locality and spatial locality in the management of the buffer cache. Traditionally only temporal locality is exploited for the purpose, and spatial locality, which refers to the on-disk sequentiality of requested blocks, is largely ignored. As the throughput of access of sequentially-placed disk blocks can be an order of magnitude higher than that of access to randomly-placed blocks, the missing of spatial locality in the buffer management can cause the performance of applications without dominant sequential accesses to be seriously degraded. In the chapter, we introduce a state-of-the-art technique that seamlessly combines these two locality properties embedded in the data access patterns into the management of the kernel buffer cache management. After elaboration on why the spatial locality is needed in addition to the temporal locality, we detail a framework, DULO (DUal LOcality), in which these two properties are taken account of simultaneously. A prototype implementation of DULO in the Linux kernel as well as some experiment results are presented, showing that DULO can significantly increases disk I/O throughput for real-world applications such as Web server, TPC benchmark, file system benchmark, and scientific programs. It reduces their execution times by as much as 53%. We conclude the chapter by identifying and encouraging a new direction for research and practice on the improvement of disk I/O performance, which is to expose more disk-specific data layout and access patterns to the upper-level system software for disk-oriented policies.

scholarly journals Application of Data Mining on Web Usage Data for Security: WebSecuDMiner

2019 ◽  
Author(s):  
Muhammad Zia Aftab Khan ◽  
Jihyun Park
Keyword(s):  
Design Methodology ◽  
Access Pattern ◽  
User Research ◽  
Web Log ◽  
Web Access ◽  
User Access ◽  
Log File ◽  
Access Patterns ◽  
Web Access Pattern ◽  
The Web

The purpose of this paper is to develop WebSecuDMiner algorithm to discover unusual web access patterns based on analysing the potential rules hidden in web server log and user navigation history. Design/methodology/approach: WebSecuDMiner uses equivalence class transformation (ECLAT) algorithm to extract user access patterns from the web log data, which will be used to identify the user access behaviours pattern and detect unusual one. Data extracted from the web serve log and user browsing behaviour is exploited to retrieve the web access pattern that is produced by the same user. Findings: WebSecuDMiner is used to detect whether any unauthorized access have been posed and take appropriate decisions regarding the review of the original rights of suspicious user. Research limitations/implications: The present work uses the database which is extracted from web serve log file and user browsing behaviour. Although the page is viewed by the user, the visit is not recorded in the server log file, since it can be access from the browser's cache.

scholarly journals Penggunaan Metode berbasis Graph untuk Mining Frequent Sequential Access Pattern Pada Studi Kasus : Website iGracias Universitas Telkom

2017 ◽  
Vol 2 (1) ◽  
pp. 91 ◽  
Author(s):  
Rahmi Rohdiniyah ◽  
Ibnu Asror ◽  
Gede Agung Ary Wisudawan
Keyword(s):  
Web Mining ◽  
Web Usage Mining ◽  
Access Pattern ◽  
Web Usage ◽  
Access Patterns

Penggunaan <em>website</em> pada bidang pendidikan, khususnya sebuah universitas, bertujuan untuk menyimpan berbagai informasi yang ada pada lingkungan universitas tersebut. Untuk itu, perlu dilakukan perbaikan struktur untuk memelihara kualitas dari web. Salah satu teknik yang dapat digunakan adalah dengan menggunakan <em>web usage mining. Web usage mining</em> merupakan salah satu cabang dari <em>web mining</em> yang digunakan untuk menemukan informasi atau pengetahuan yang bermanfaat dari pola navigasi <em>user </em>pada sebuah<em> website</em>. Pada penelitian ini menggunakan metode berbasis graph untuk <em>frequent sequential access patterns</em> dan menggunakan Igracias Universitas Telkom sebagai studi kasusnya. Karena Igracias selalu digunakan oleh seluruh entitas yang ada pada Universitas Telkom. Metode ini  memiliki kelebihan untuk menemukan <em>behavior</em> pola pengaksesan <em>user</em>. Dari implementasi metoda ini didapat pola akses group user secara berurutan.

scholarly journals Flat ORAM: A Simplified Write-Only Oblivious RAM Construction for Secure Processors

Cryptography ◽  
2019 ◽  
Vol 3 (1) ◽  
pp. 10 ◽  
Author(s):  
Syed Haider ◽  
Marten van Dijk
Keyword(s):  
Energy Savings ◽  
Performance Gain ◽  
Access Pattern ◽  
Performance Bottlenecks ◽  
Oblivious Ram ◽  
Privacy Leakage ◽  
Cryptographic Primitive ◽  
Secure Processors ◽  
Performance Penalty ◽  
Access Patterns

Oblivious RAM (ORAM) is a cryptographic primitive which obfuscates the access patterns to a storage, thereby preventing privacy leakage. So far in the current literature, only ‘fully functional’ ORAMs are widely studied which can protect, at a cost of considerable performance penalty, against the strong adversaries who can monitor all read and write operations. However, recent research has shown that information can still be leaked even if only the write access pattern (not reads) is visible to the adversary. For such weaker adversaries, a fully functional ORAM turns out to be an overkill, causing unnecessary overheads. Instead, a simple ‘write-only’ ORAM is sufficient, and, more interestingly, is preferred as it can offer far better performance and energy efficiency than a fully functional ORAM. In this work, we present Flat ORAM: an efficient write-only ORAM scheme which outperforms the closest existing write-only ORAM called HIVE. HIVE suffers from performance bottlenecks while managing the memory occupancy information vital for correctness of the protocol. Flat ORAM introduces a simple idea of Occupancy Map (OccMap) to efficiently manage the memory occupancy information resulting in far better performance. Our simulation results show that, compared to HIVE, Flat ORAM offers 50 % performance gain on average and up to 80 % energy savings.

scholarly journals A replacement strategy for a distributed caching system based on the spatiotemporal access pattern of geospatial data

2014 ◽  
Vol XL-4 ◽  
pp. 133-137 ◽  
Author(s):  
R. Li ◽  
X. Wang ◽  
X. Shi
Keyword(s):  
Data Access ◽  
Geospatial Data ◽  
Combination Method ◽  
System Throughput ◽  
Time Interval ◽  
Cache Replacement ◽  
Access Pattern ◽  
Replacement Strategy ◽  
Hit Rate ◽  
Access Patterns

Cache replacement strategy is the core for a distributed high-speed caching system, and effects the cache hit rate and utilization of a limited cache space directly. Many reports show that there are temporal and spatial local changes in access patterns of geospatial data, and there are popular hot spots which change over time. Therefore, the key issue for cache replacement strategy for geospatial data is to get a combination method which considers both temporal local changes and spatial local changes in access patterns, and balance the relationship between the changes. And the cache replacement strategy should fit the distribution and changes of hotspot. This paper proposes a cache replacement strategy based on access pattern which have access spatiotemporal localities. Firstly, the strategy builds a method to express the access frequency and the time interval for geospatial data access based on a least-recently-used replacement (LRU) algorithm and its data structure; secondly, considering both the spatial correlation between geospatial data access and the caching location for geospatial data, it builds access sequences based on a LRU stack, which reflect the spatiotemporal locality changes in access pattern. Finally, for achieving the aim of balancing the temporal locality and spatial locality changes in access patterns, the strategy chooses the replacement objects based on the length of access sequences and the cost of caching resource consumption. Experimental results reveal that the proposed cache replacement strategy is able to improve the cache hit rate while achieving a good response performance and higher system throughput. Therefore, it can be applied to handle the intensity of networked GISs data access requests in a cloud-based environment.

scholarly journals Data Mining for Information Storage Reliability Assessment by Relative Values

2018 ◽  
Vol 7 (4.7) ◽  
pp. 204 ◽  
Author(s):  
Iskandar N. Nasyrov ◽  
Ildar I. Nasyrov ◽  
Rustam I. Nasyrov ◽  
Bulat A. Khairullin
Keyword(s):  
Data Storage ◽  
Error Rate ◽  
Hard Disk Drives ◽  
Hard Disk ◽  
Careful Analysis ◽  
Information Storage ◽  
Storage Device ◽  
Relative Distribution ◽  
Solid State Drives ◽  
Reliability Indicators

The data ambiguity problem for heterogeneous sets of equipment reliability indicators is considered. In fact, the same manufacturers do not always unambiguously fill the SMART parameters with the corresponding values for their different models of hard disk drives. In addition, some of the parameters are sometimes empty, while the other parameters have only zero values.The scientific task of the research consists in the need to define such a set of parameters that will allow us to obtain a comparative assessment of the reliability of each individual storage device of any model of any manufacturer for its timely replacement.The following conditions were used to select the parameters suitable for evaluating their relative values:1) The parameter values for normally operating drives should always be greater or lower than for the failed ones;2) The monotonicity of changes in the values of parameters in the series should be observed: normally working, withdrawn prematurely, failed;3) The first two conditions must be fulfilled both in general and in particular, for example, for the drives of each brand separately.Separate averaging of the values for normally operating, early decommissioned and failed storage media was performed. The maximum of these three values was taken as 100%. The relative distribution of values for each parameter was studied.Five parameters were selected (5 – “Reallocated sectors count”, 7 – “Seek error rate”, 184 – “End-to-end error”, 196 – “Reallocation event count”, 197 – “Current pending sector count”, plus another four (1 – “Raw read error rate”, 10 – “Spin-up retry counts”, 187 – “Reported uncorrectable errors”, 198 – “Uncorrectable sector counts”), which require more careful analysis, and one (194 – “Hard disk assembly temperature”) for prospective use in solid-state drives, as a result of the relative value study of their suitability for use upon evaluating the reliability of data storage devices. 

scholarly journals Efficient Use of GPU Memory for Large-Scale Deep Learning Model Training

2021 ◽  
Vol 11 (21) ◽  
pp. 10377
Author(s):  
Hyeonseong Choi ◽  
Jaehwan Lee
Keyword(s):  
Deep Learning ◽  
Large Scale ◽  
Training Model ◽  
High Accuracy ◽  
Access Pattern ◽  
Learning Models ◽  
Training Models ◽  
Model Training ◽  
Access Patterns ◽  
Deep Learning Model

To achieve high accuracy when performing deep learning, it is necessary to use a large-scale training model. However, due to the limitations of GPU memory, it is difficult to train large-scale training models within a single GPU. NVIDIA introduced a technology called CUDA Unified Memory with CUDA 6 to overcome the limitations of GPU memory by virtually combining GPU memory and CPU memory. In addition, in CUDA 8, memory advise options are introduced to efficiently utilize CUDA Unified Memory. In this work, we propose a newly optimized scheme based on CUDA Unified Memory to efficiently use GPU memory by applying different memory advise to each data type according to access patterns in deep learning training. We apply CUDA Unified Memory technology to PyTorch to see the performance of large-scale learning models through the expanded GPU memory. We conduct comprehensive experiments on how to efficiently utilize Unified Memory by applying memory advises when performing deep learning. As a result, when the data used for deep learning are divided into three types and a memory advise is applied to the data according to the access pattern, the deep learning execution time is reduced by 9.4% compared to the default Unified Memory.

scholarly journals SqORAM: Read-Optimized Sequential Write-Only Oblivious RAM

2020 ◽  
Vol 2020 (1) ◽  
pp. 216-234
Author(s):  
Anrin Chakraborti ◽  
Radu Sion
Keyword(s):  
Random Access ◽  
Data Access ◽  
Data Locality ◽  
Storage Device ◽  
Random Data ◽  
Locality Property ◽  
Oblivious Ram ◽  
Locality Preserving ◽  
Access Patterns ◽  
Access Data

AbstractOblivious RAMs (ORAMs) allow a client to access data from an untrusted storage device without revealing the access patterns. Typically, the ORAM adversary can observe both read and write accesses. Write-only ORAMs target a more practical, multi-snapshot adversary only monitoring client writes – typical for plausible deniability and censorship-resilient systems. This allows write-only ORAMs to achieve significantly-better asymptotic performance. However, these apparent gains do not materialize in real deployments primarily due to the random data placement strategies used to break correlations between logical and physical names-paces, a required property for write access privacy. Random access performs poorly on both rotational disks and SSDs (often increasing wear significantly, and interfering with wear-leveling mechanisms).In this work, we introduce SqORAM, a new locality-preserving write-only ORAM that preserves write access privacy without requiring random data access. Data blocks close to each other in the logical domain land in close proximity on the physical media. Importantly, SqORAM maintains this data locality property over time, significantly increasing read throughput.A full Linux kernel-level implementation of SqORAM is 100x faster than non locality-preserving solutions for standard workloads and is 60-100% faster than the state-of-the-art for typical file system workloads.

Sign in / Sign up

Export Citation Format

Share Document