We propose an I/O classification architecture to close the widening semantic gap between computer systems and storage systems. By classifying I/O, a computer system can request that different classes of data be handled with different storage system policies. Specifically, when a storage system is first initialized, we assign performance policies to predefined classes, such as the filesystem journal. Then, online, we include a classifier with each I/O command (e.g., SCSI), thereby allowing the storage system to enforce the associated policy for each I/O that it receives. Our immediate application is caching. We present filesystem prototypes and a database proof-of-concept that classify all disk I/O — with very little modification to the filesystem, database, and operating system. We associate caching policies with various classes (e.g., large files shall be evicted before metadata and small files), and we show that end-to-end file system performance can be improved by over a factor of two, relative to conventional caches like LRU. And caching is simply one of many possible applications. As part of our ongoing work, we are exploring other classes, policies and storage system mechanisms that can be used to improve end-to-end performance, reliability and security.

Flash Memory based Solid State Drive (SSD) has been called a “pivotal technology ” that could revolutionize data storage systems. Since SSD shares a common interface with the traditional hard disk drive (HDD), both physically and logically, an effective integration of SSD into the storage hierarchy is very important. However, details of SSD hardware implementations tend to be hidden behind such narrow interfaces. In fact, since sophisticated algorithms are usually, of necessity, adopted in SSD controller firmware, more complex performance dynamics are to be expected in SSD than in HDD systems. Most existing literature or product specifications on SSD just provide high-level descriptions and standard performance data, such as bandwidth and latency. In order to gain insight into the unique performance characteristics

Abstract—Parallel applications are usually able to achieve high computational performance but suffer from large latency in I/O accesses. I/O prefetching is an effective solution for masking the latency. Most of existing I/O prefetching techniques, however, are conservative and their effectiveness is limited by low accuracy and coverage. As the processor-I/O performance gap has been increasing rapidly, data-access delay has become a dominant performance bottleneck. We argue that it is time to revisit the “I/O wall ” problem and trade the excessive computing power with data-access speed. We propose a novel pre-execution approach for masking I/O latency. We describe the pre-execution I/O prefetching framework, the pre-execution thread construction methodology, the underlying library support, and the prototype implementation in the ROMIO MPI-IO implementation in MPICH2. Preliminary experiments show that the pre-execution approach is promising in reducing I/O access latency and has real potential. I.

State-of-the-art networked storage servers are equipped with increasingly powerful computing capability and large DRAM memory as storage caches. However, their contribution to the performance improvement of networked storage system has become increasingly limited. This is because the client-side memory sizes are also increasing, which reduces capacity misses in the client buffer caches as well as access locality in the storage servers, thus weakening the caching effectiveness of server storage caches. Proactive caching in storage servers is highly desirable to reduce cold misses in clients. We propose an effective way to improve the utilization of storage server resources through prefetching in storage servers for clients. In particular, our design well utilizes two unique strengths of networked storage servers which are not leveraged in existing storage server prefetching schemes. First, powerful storage servers have idle CPU cycles, under-utilized disk bandwidth, and abundant memory space, providing many opportunities for aggressive disk data prefetching. Second, the servers have the knowledge about high-latency operations in storage devices, such as disk head positioning, which enables efficient disk data prefetching based on an accurate cost-benefit analysis of prefetch operations. We present STEP – a Sequentiality and Thrashing dEtection based Prefetching scheme, and its implementation with Linux Kernel 2.6.16. Our performance evaluation by replaying Storage Performance Council (SPC)’s OLTP traces shows that server performance improvements are up to 94% with an average of 25%. Improvements with frequently used Unix applications are up to 53 % with an average of 12%. Our experiments also show that STEP has little effect on workloads with random access patterns, such as SPC ’ Web-Search traces. 1

Memory trace analysis is an important technology for architecture research, system software (i.e., OS, compiler) optimization, and application performance improvements. Many approaches have been used to track memory trace, such as simulation, binary instrumentation and hardware snooping. However, they usually have limitations of time, accuracy and capacity. In this paper we propose a platform independent memory trace monitoring system, which is able to track virtual memory reference trace of full systems (including OS, VMMs, libraries, and applications). The system adopts a DIMM-snooping mechanism that uses hardware boards plugged in DIMM slots to snoop. There are several advantages in this approach, such as fast, complete, undistorted, and portable. Three key techniques are proposed to address the system design challenges with this

Abstract—Power is the greatest lifetime cost in an archival system, and, as decreasing costs make disks more attractive than tapes, spinning disks account for the majority of power drawn. To reduce this cost, we propose reducing the number of times disks have to spin up by grouping together files such that a typical spin-up handles several file accesses. For a typical system, we show that if only 30 % of total accesses occur while disks are still spinning, we can conserve 12 % of the power cost. We classify files according to directory structure and see access hit rates of up to 66 % for a power savings of up to 52 % of the power cost of spinning up for every read in easily-separable workloads. I.

On-disk sequentiality of requested blocks, or their spatial locality, is critical to real disk performance where the throughput of access to sequentially-placed disk blocks can be an order of magnitude higher than that of access to randomly-placed blocks. Unfortunately, spatial locality of cached blocks is largely ignored, and only temporal locality is considered in current system buffer cache managements. Thus, disk performance for workloads without dominant sequential accesses can be seriously degraded. To address this problem, we propose a scheme called DULO (DUal LOcality) which exploits both temporal and spatial localities in the buffer cache management. Leveraging the filtering effect of the buffer cache, DULO can influence the I/O request stream by making the requests passed to the disk more sequential, thus significantly increasing the effectiveness of I/O scheduling and prefetching for disk performance improvements. We have implemented a prototype of DULO in Linux 2.6.11. The implementation shows that DULO can significantly increases disk I/O throughput for real-world applications such as a Web server, TPC benchmark, file system benchmark, and scientific programs. It reduces their execution times by as much as 53%.

Extension of battery lifetime has always been a major issue for mobile computing. While more and more data are involved in mobile computing, energy consumption caused by I/O operations becomes increasingly large. In a pervasive computing environment, the requested data can be stored both on the local disk of a mobile computer by using the hoarding technique, and on the remote server, where data are accessible via wireless communication. Based on the current operational states of local disk (active or standby), the amount of data to be requested (small or large), and currently available wireless bandwidth (strong or weak reception), data access source can be adaptively selected to achieve maximum energy reduction. To this end, we propose a profile-based I/O management scheme, FlexFetch, that is aware of access history and adaptive to current access environment. Our simulation experiments driven by real-life traces demonstrate that the scheme can significantly reduce energy consumption in a mobile computer compared with existing representative schemes. 1

The analysis of data usage in a large set of real traces from a highenergy physics collaboration revealed the existence of an emergent grouping of files that we coined “filecules”. This paper presents the benefits of using this file grouping for prestaging data and compares it with previously proposed file grouping techniques along a range of performance metrics. Our experiments with real workloads demonstrate that filecule grouping is a reliable and useful abstraction for data management in science Grids; that preserving time locality for data prestaging is highly recommended; that job reordering with respect to data availability has significant impact on throughput; and finally, that a relatively short history of traces is a good predictor for filecule grouping. Our experimental results provide lessons for workload modeling and suggest design guidelines for data management in dataintensive resource-sharing environments.

The multi-level storage architecture has been widely adopted in servers and data centers. However, while prefetching has been shown as a crucial technique to exploit the sequentiality in accesses common for such systems and hide the increasing relative cost of disk I/O, existing multi-level storage studies have focused mostly on cache replacement strategies. In this paper, we show that prefetching algorithms designed for single-level systems may have their limitations magnified when applied to multi-level systems. Overly conservative prefetching will not be able to effectively use the lower-level cache space, while overly aggressive prefetching will be compounded across levels and generate large amounts of wasted prefetch. We take an innovative approach to this problem: rather than designing