The Experts below are selected from a list of 213 Experts worldwide ranked by ideXlab platform
Gregory R. Ganger - One of the best experts on this subject based on the ideXlab platform.
-
MultiMap: Preserving Disk Locality for Multidimensional Datasets (CMU-PDL-05-102)
2018Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Christos Faloutsos, Gregory R. GangerAbstract:MultiMap is a new approach to mapping multidimensional datasets to the linear address space of storage systems. MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap’s superior performance for range and beam queries. On average, MultiMap reduces overall I/O time by over 50% when compared to traditional naive layouts and by over 30% when compared to a Hilbert curve approach. For scans of the primary dimension, MultiMap and naive both provide almost two orders of magnitude higher throughput than the Hilbert curve approach
-
Abstract Track-aligned Extents: Matching Access Patterns to Disk Drive Characteristics
2008Co-Authors: Jiri Schindler, Christopher R. Lumb, John Linwood Griffin, Gregory R. GangerAbstract:Track-aligned extents (traxtents) utilize disk-specific knowledge to match access patterns to the strengths of modern disks. By allocating and accessing related data on disk track boundaries, a system can avoid most Rotational Latency and track crossing overheads. Avoiding these overheads can increase disk access efficiency by up to 50 % for mid-sized requests (100–500 KB). This paper describes traxtents, algorithms for detecting track boundaries, and some uses of traxtents in file systems and video servers. For large-file workloads, a version of FreeBSD’s FFS implementation that exploits traxtents reduces application run times by up to 20 % compared to the original version. A video server using traxtent-based requests can support 56 % more concurrent streams at the same startup Latency and buffer space. For LFS, 44 % lower overall write cost for track-sized segments can be achieved
-
ICDE - MultiMap: Preserving disk locality for multidimensional datasets
2007 IEEE 23rd International Conference on Data Engineering, 2007Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Gregory R. GangerAbstract:MultiMap is an algorithm for mapping multidimensional datasets so as to preserve the data's spatial locality on disks. Without revealing disk-specific details to applications, MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap's superior performance for range and beam queries. On average, MultiMap reduces total I/O time by over 50% when compared to traditional linearized layouts and by over 30% when compared to space-filling curve approaches such as Z-ordering and Hilbert curves. For scans of the primary dimension, MultiMap and traditional linearized layouts provide almost two orders of magnitude higher throughput than space-filling curve approaches.
-
MultiMap: Preserving Disk Locality for Multidimensional Datasets (CMU-PDL-05-102)
2005Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Christos Faloutsos, Gregory R. GangerAbstract:MultiMap is a new approach to mapping multidimensional datasets to the linear address space of storage systems. MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap’s superior performance for range and beam queries. On average, MultiMap reduces overall I/O time by over 50% when compared to traditional naive layouts and by over 30% when compared to a Hilbert curve approach. For scans of the primary dimension, MultiMap and naive both provide almost two orders of magnitude higher throughput than the Hilbert curve approach. We thank the members and companies of the PDL Consortium (including EMC, Hewlett-Packard, Hitachi, IBM, Intel, Microsoft, Network Appliance, Oracle, Panasas, Seagate, Sun, and Veritas) for their interest, insights, feedback, and support. This work is funded in part by NSF grants CCR-0113660, IIS-0133686, and CCR-0205544, as well as by an IBM faculty partnership award.
-
FAST - Freeblock scheduling outside of disk firmware
2002Co-Authors: Christopher R. Lumb, Jiri Schindler, Gregory R. GangerAbstract:Freeblock scheduling replaces a disk drive's Rotational Latency delays with useful background media transfers, potentially allowing background disk I/O to occur with no impact on foreground service times. To do so, a freeblock scheduler must be able to very accurately predict the service time components of any given disk request -- the necessary accuracy was not previously considered achievable outside of disk firmware. This paper describes the design and implementation of a working external freeblock scheduler running either as a user-level application atop Linux or inside the FreeBSD kernel. This freeblock scheduler can give 15% of a disk's potential bandwidth (over 3.1MB/s) to a background disk scanning task with almost no impact (less than 2%) on the foreground request response times. This can increase disk bandwidth utilization by over 6×.
Kyu Ho Park - One of the best experts on this subject based on the ideXlab platform.
-
Arrangement of multi-dimensional scalable video data for heterogeneous clients
Information Systems, 2010Co-Authors: Man-keun Seo, Sung Hoon Baek, Kyu Ho ParkAbstract:A scalable video server extracts data corresponding to the resolution requested by its client from the total data containing the information encoding a full resolution video. Depending on the requested resolution, the extracted data may not be contiguously placed on a disk or a disk array. For this reason, the traverse distance, which indicates the difference between the first read position and the last read position, can be much larger than the amount of the requested data. This causes additional Rotational Latency in a disk and thus degrades disk performance. Furthermore, scalable video data more seriously deteriorates the independency of disks in a disk array. That is, even a small read request can be split into multiple disk requests across disks of a disk array, because the requested data are scattered across multiple disks. To address these problems, we propose new data arrangement schemes for scalable video data. In these new schemes, we first deal with the arrangement of multi-dimensional scalable video data, which can be employed regardless of the number of scalability dimensions. Second, we improve disk performance by reducing average disk cost, which is based on both the traverse distance of each disk and the independency of disks. Third, we improve overall performance of disk devices through considering the entire request pattern, when large numbers of clients concurrently demand heterogeneous resolutions of videos from a server. We also propose fast arrangement algorithms to reduce the computation time required for searching an effective arrangement so that they can be easily applied to practical server system.
-
A New RAID-Disk Placement Method for Interactive Media Server with an Accurate Bit Count Control
2008Co-Authors: Yo-won Jeong, Seung-ho Lim, Kyu Ho ParkAbstract:Abstract. In this paper, we propose a RAID-disk placement algorithm of coded video data and an efficient disk prefetching method to increase the number of clients who can be serviced interactive operations in the media server. Our placement policy is incorporated with a special bit count control method that is based on repeated tuning of quantization parameters to adjust the actual bit count to the target bit count. The encoder using this method can generate coded frames whose sizes are synchronized with the RAID stripe size, so that when various fast-forward levels are accessed we can reduce the seek and Rotational Latency and enhance the disk throughput.
-
Data placement and prefetching with accurate bit rate control for interactive media server
ACM Transactions on Multimedia Computing Communications and Applications, 2008Co-Authors: Seung-ho Lim, Yo-won Jeong, Kyu Ho ParkAbstract:An interactive Media Server should support unrestricted control to viewers with their service level agreements. It is important to manage video data effectively to facilitate efficient retrieval. In this paper, we propose an efficient placement algorithm as part of an effective retrieval scheme to increase the number of clients who can be provided with interactive service. The proposed management schemes are incorporated with a bit count control method that is based on repeated tuning of quantization parameters to adjust the actual bit count to the target bit count. The encoder using this method can generate coded frames whose sizes are synchronized with the RAID stripe size, so that when various fast-forward levels are accessed we can reduce the seek and Rotational Latency and enhance the disk throughput of each disk in the RAID system. Experimental results demonstrate that the proposed schemes can significantly improve the average service time and guarantee more users service of quality, and the interactive media server can thereby efficiently service a large number of clients.
-
NPC - A new raid-disk placement method for interactive media server with an accurate bit count control
Lecture Notes in Computer Science, 2005Co-Authors: Yo-won Jeong, Seung-ho Lim, Kyu Ho ParkAbstract:In this paper, we propose a RAID-disk placement algorithm of coded video data and an efficient disk prefetching method to increase the number of clients who can be serviced interactive operations in the media server. Our placement policy is incorporated with a special bit count control method that is based on repeated tuning of quantization parameters to adjust the actual bit count to the target bit count. The encoder using this method can generate coded frames whose sizes are synchronized with the RAID stripesize, so that when various fast-forward levels are accessed we can reduce the seek and Rotational Latency and enhance the disk throughput.
-
NOSSDAV - Interactive media server with media synchronized RAID storage system
Proceedings of the international workshop on Network and operating systems support for digital audio and video - NOSSDAV '05, 2005Co-Authors: Seung-ho Lim, Yo-won Jeong, Kyu Ho ParkAbstract:We propose an efficient placement algorithm and per-disk prefetching method to effectively support interactive operations in the media server. Our placement policy is incorporated with an encoder having a special bitcount control scheme that repeatedly tunes quantization parameters to adjust the bitcounts of video frames. This encoder can generate coded frames sub-stream video blocks whose sizes are synchronized with the RAID stripe size, so that when various fast-forward levels are accessed we can reduce the seek and Rotational Latency and enhance the disk throughput of each disk in the RAID system. In the experimental results, the proposed placement policy and bitrate control scheme can significantly improve the average service time, which can enlarge the capacity of the interactive media server.
Heonshik Shin - One of the best experts on this subject based on the ideXlab platform.
-
new disk scheduling algorithms for reduced Rotational Latency
Database Systems for Advanced Applications, 1993Co-Authors: Kitae Hwang, Heonshik ShinAbstract:Disk arm scheduling algorithms have been studied for many years to increase disk I/O performance. Most of the disks by the early 1980’s are characterized as a linear seek time and their seek time is responsible for the most time of disk access. So, the existing disk scheduling algorithms have focused on the reduction of the average seek distance. Seek time has improved greatly and today’s disks usually have nonlinear seek time characteristics, whereas the rotation speed has been steady at the 3600 RPM. So, it is of value to reduce average Rotational Latency. In this paper we propose two disk scheduling algorithms, i.e. Shortest Rotational Latency First(SRLF) and Shortest Access Tie Fist(SATF) for reduced Rotational Latency. SRLF services first a request with the shortest expected Rotational Latency and SATF services first a request with the shortest expected access time. We analyze the expected access time of the request serviced first by each of SSTF and SRLF under the uniform cylinder access. We evaluated the response time of SSTF, SRLF, and SATF through the simulations under the uniform and localized cylinder access. The results of the analysis and simulations show that the algorithms proposed in this paper are more efficient than SSTF which is known for the fastest response time.
-
DASFAA - New Disk Scheduling Algorithms for Reduced Rotational Latency
Database Systems for Advanced Applications '93, 1993Co-Authors: Kitae Hwang, Heonshik ShinAbstract:Disk arm scheduling algorithms have been studied for many years to increase disk I/O performance. Most of the disks by the early 1980’s are characterized as a linear seek time and their seek time is responsible for the most time of disk access. So, the existing disk scheduling algorithms have focused on the reduction of the average seek distance. Seek time has improved greatly and today’s disks usually have nonlinear seek time characteristics, whereas the rotation speed has been steady at the 3600 RPM. So, it is of value to reduce average Rotational Latency. In this paper we propose two disk scheduling algorithms, i.e. Shortest Rotational Latency First(SRLF) and Shortest Access Tie Fist(SATF) for reduced Rotational Latency. SRLF services first a request with the shortest expected Rotational Latency and SATF services first a request with the shortest expected access time. We analyze the expected access time of the request serviced first by each of SSTF and SRLF under the uniform cylinder access. We evaluated the response time of SSTF, SRLF, and SATF through the simulations under the uniform and localized cylinder access. The results of the analysis and simulations show that the algorithms proposed in this paper are more efficient than SSTF which is known for the fastest response time.
Jiri Schindler - One of the best experts on this subject based on the ideXlab platform.
-
MultiMap: Preserving Disk Locality for Multidimensional Datasets (CMU-PDL-05-102)
2018Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Christos Faloutsos, Gregory R. GangerAbstract:MultiMap is a new approach to mapping multidimensional datasets to the linear address space of storage systems. MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap’s superior performance for range and beam queries. On average, MultiMap reduces overall I/O time by over 50% when compared to traditional naive layouts and by over 30% when compared to a Hilbert curve approach. For scans of the primary dimension, MultiMap and naive both provide almost two orders of magnitude higher throughput than the Hilbert curve approach
-
Abstract Track-aligned Extents: Matching Access Patterns to Disk Drive Characteristics
2008Co-Authors: Jiri Schindler, Christopher R. Lumb, John Linwood Griffin, Gregory R. GangerAbstract:Track-aligned extents (traxtents) utilize disk-specific knowledge to match access patterns to the strengths of modern disks. By allocating and accessing related data on disk track boundaries, a system can avoid most Rotational Latency and track crossing overheads. Avoiding these overheads can increase disk access efficiency by up to 50 % for mid-sized requests (100–500 KB). This paper describes traxtents, algorithms for detecting track boundaries, and some uses of traxtents in file systems and video servers. For large-file workloads, a version of FreeBSD’s FFS implementation that exploits traxtents reduces application run times by up to 20 % compared to the original version. A video server using traxtent-based requests can support 56 % more concurrent streams at the same startup Latency and buffer space. For LFS, 44 % lower overall write cost for track-sized segments can be achieved
-
ICDE - MultiMap: Preserving disk locality for multidimensional datasets
2007 IEEE 23rd International Conference on Data Engineering, 2007Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Gregory R. GangerAbstract:MultiMap is an algorithm for mapping multidimensional datasets so as to preserve the data's spatial locality on disks. Without revealing disk-specific details to applications, MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap's superior performance for range and beam queries. On average, MultiMap reduces total I/O time by over 50% when compared to traditional linearized layouts and by over 30% when compared to space-filling curve approaches such as Z-ordering and Hilbert curves. For scans of the primary dimension, MultiMap and traditional linearized layouts provide almost two orders of magnitude higher throughput than space-filling curve approaches.
-
MultiMap: Preserving Disk Locality for Multidimensional Datasets (CMU-PDL-05-102)
2005Co-Authors: Minglong Shao, Jiri Schindler, Steven W. Schlosser, Stratos Papadomanolakis, Anastassia Ailamaki, Christos Faloutsos, Gregory R. GangerAbstract:MultiMap is a new approach to mapping multidimensional datasets to the linear address space of storage systems. MultiMap exploits modern disk characteristics to provide full streaming bandwidth for one (primary) dimension and maximally efficient non-sequential access (i.e., minimal seek and no Rotational Latency) for the other dimensions. This is in contrast to existing approaches, which either severely penalize non-primary dimensions or fail to provide full streaming bandwidth for any dimension. Experimental evaluation of a prototype implementation demonstrates MultiMap’s superior performance for range and beam queries. On average, MultiMap reduces overall I/O time by over 50% when compared to traditional naive layouts and by over 30% when compared to a Hilbert curve approach. For scans of the primary dimension, MultiMap and naive both provide almost two orders of magnitude higher throughput than the Hilbert curve approach. We thank the members and companies of the PDL Consortium (including EMC, Hewlett-Packard, Hitachi, IBM, Intel, Microsoft, Network Appliance, Oracle, Panasas, Seagate, Sun, and Veritas) for their interest, insights, feedback, and support. This work is funded in part by NSF grants CCR-0113660, IIS-0133686, and CCR-0205544, as well as by an IBM faculty partnership award.
-
FAST - Freeblock scheduling outside of disk firmware
2002Co-Authors: Christopher R. Lumb, Jiri Schindler, Gregory R. GangerAbstract:Freeblock scheduling replaces a disk drive's Rotational Latency delays with useful background media transfers, potentially allowing background disk I/O to occur with no impact on foreground service times. To do so, a freeblock scheduler must be able to very accurately predict the service time components of any given disk request -- the necessary accuracy was not previously considered achievable outside of disk firmware. This paper describes the design and implementation of a working external freeblock scheduler running either as a user-level application atop Linux or inside the FreeBSD kernel. This freeblock scheduler can give 15% of a disk's potential bandwidth (over 3.1MB/s) to a background disk scanning task with almost no impact (less than 2%) on the foreground request response times. This can increase disk bandwidth utilization by over 6×.
Kitae Hwang - One of the best experts on this subject based on the ideXlab platform.
-
new disk scheduling algorithms for reduced Rotational Latency
Database Systems for Advanced Applications, 1993Co-Authors: Kitae Hwang, Heonshik ShinAbstract:Disk arm scheduling algorithms have been studied for many years to increase disk I/O performance. Most of the disks by the early 1980’s are characterized as a linear seek time and their seek time is responsible for the most time of disk access. So, the existing disk scheduling algorithms have focused on the reduction of the average seek distance. Seek time has improved greatly and today’s disks usually have nonlinear seek time characteristics, whereas the rotation speed has been steady at the 3600 RPM. So, it is of value to reduce average Rotational Latency. In this paper we propose two disk scheduling algorithms, i.e. Shortest Rotational Latency First(SRLF) and Shortest Access Tie Fist(SATF) for reduced Rotational Latency. SRLF services first a request with the shortest expected Rotational Latency and SATF services first a request with the shortest expected access time. We analyze the expected access time of the request serviced first by each of SSTF and SRLF under the uniform cylinder access. We evaluated the response time of SSTF, SRLF, and SATF through the simulations under the uniform and localized cylinder access. The results of the analysis and simulations show that the algorithms proposed in this paper are more efficient than SSTF which is known for the fastest response time.
-
DASFAA - New Disk Scheduling Algorithms for Reduced Rotational Latency
Database Systems for Advanced Applications '93, 1993Co-Authors: Kitae Hwang, Heonshik ShinAbstract:Disk arm scheduling algorithms have been studied for many years to increase disk I/O performance. Most of the disks by the early 1980’s are characterized as a linear seek time and their seek time is responsible for the most time of disk access. So, the existing disk scheduling algorithms have focused on the reduction of the average seek distance. Seek time has improved greatly and today’s disks usually have nonlinear seek time characteristics, whereas the rotation speed has been steady at the 3600 RPM. So, it is of value to reduce average Rotational Latency. In this paper we propose two disk scheduling algorithms, i.e. Shortest Rotational Latency First(SRLF) and Shortest Access Tie Fist(SATF) for reduced Rotational Latency. SRLF services first a request with the shortest expected Rotational Latency and SATF services first a request with the shortest expected access time. We analyze the expected access time of the request serviced first by each of SSTF and SRLF under the uniform cylinder access. We evaluated the response time of SSTF, SRLF, and SATF through the simulations under the uniform and localized cylinder access. The results of the analysis and simulations show that the algorithms proposed in this paper are more efficient than SSTF which is known for the fastest response time.
