Flash in a DBMS: Where and How?
Published 1 January 2010Open access
Manos Athanassoulis, Anastasia Ailamaki, Shimin Chen, Phillip B. Gibbons, Radu Stoica
Citations25
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
Techniques for making effective use of flash in three contexts are described: as a log device for transaction processing on memory-resident data, as the main data store for transactionprocessing, and as an update cache for HDD- resident data warehouses.
Abstract
Over the past decade, new solid state storage technologies, with flash being the most mature one, have become increasingly popular. Such technologies store data durably, and can alleviate many handicaps of hard disk drives (HDDs). Nonetheless, they have very different characteristics compared to HDDs, making it challenging to integrate such technologies into data intensive systems, such as database management systems (DBMS), that rely heavily on underlying storage behaviors. In this paper, we ask the question: Where and how will flash be exploited in a DBMS? We describe techniques for making effective use of flash in three contexts: (i) as a log device for transaction processing on memory-resident data, (ii) as the main data store for transaction processing, and (iii) as an update cache for HDD-resident data warehouses.
Keywords
Computer Science
Computer Architecture: A Quantitative Approach
9,536 Citations1989John L. Hennessy, David A. Patterson
This best-selling title, considered for over a decade to be essential reading for every serious student and practitioner of computer design, has been updated throughout to address the most important trends facing computer designers today.
C-store: a column-oriented DBMS
923 Citations2005Mike Stonebraker, Daniel J. Abadi +12 more
Design tradeoffs for SSD performance
882 Citations2008Nitin Agrawal, Vijayan Prabhakaran +4 more
It is found that SSD performance and lifetime is highly workload-sensitive, and that complex systems problems that normally appear higher in the storage stack, or even in distributed systems, are relevant to device firmware.
IBM Journal of Research and DevelopmentOverview of candidate device technologies for storage-class memory
669 Citations2008Geoffrey W. Burr, B. N. Kurdi +4 more
This work discusses evolutionary extensions of conventional flash memory, such as SONOS and nanotraps, as well as a number of revolutionary new memory technologies, including ferroelectric, magnetic, phase-change, and resistive random-access memories, including perovskites and solid electrolytes, and finally organic and polymeric memory.
A case for flash memory ssd in enterprise database applications
299 Citations2008Sang-Won Lee, Bongki Moon +3 more
It is shown empirically that up to more than an order of magnitude improvement can be achieved in transaction processing by replacing magnetic disk with flash memory SSD for transaction log, rollback segments, and temporary table spaces.
Proceedings of the VLDB EndowmentFlashStore
239 Citations2010Biplob Debnath, Sudipta Sengupta +1 more
This work uses real-world data traces from two data center applications, namely, Xbox LIVE Primetime online multi-player game and inline storage deduplication, to drive and evaluate the design of FlashStore on traditional and low power server platforms.
Rethinking Database Algorithms for Phase Change Memory
225 Citations2011Shimin Chen, Phillip B. Gibbons +1 more
Improved algorithms that reduce both execution time and energy on PCM while increasing write endurance are presented, and current approaches for common database algorithms such as B + -trees and Hash Joins are suboptimal for PCM.
Research Portal (King's College London)uFLIP: Understanding Flash IO Patterns
153 Citations2009Luc Bouganim, Björn Þór Jónsson +1 more
This paper defines uFLIP, a benchmark for measuring the response time of flash IO patterns, and presents a benchmarking methodology which takes into account the particular characteristics of flash devices.
Query processing techniques for solid state drives
137 Citations2009Dimitris Tsirogiannis, Stavros Harizopoulos +3 more
This paper investigates data structures and algorithms that leverage fast random reads to speed up selection, projection, and join operations in relational query processing, and introduces FlashJoin, a general pipelined join algorithm that minimizes accesses to base and intermediate relational data.
Proceedings of the VLDB EndowmentSSD bufferpool extensions for database systems
133 Citations2010Mustafa Canim, George A. Mihaila +3 more
This work presents a technique for using solid-state storage as a caching layer between RAM and hard disks in database management systems by caching data that is accessed frequently, disk I/O is reduced.
Proceedings of the VLDB EndowmentLazy-Adaptive Tree
132 Citations2009Devesh Agrawal, Deepak Ganesan +3 more
The Lazy-Adaptive Tree (LA-Tree) is presented, a novel index structure that is designed to improve performance by minimizing accesses to flash by amortizing the cost of node reads and writes by performing update operations in a lazy manner using cascaded buffers.
Advances in flash memory SSD technology for enterprise database applications
125 Citations2009Sang-Won Lee, Bongki Moon +1 more
It is demonstrated that even a single flash memory drive can outperform a level-0 RAID with eight enterprise class 15k-RPM disk drives with respect to transaction throughput, cost effectiveness and energy consumption.
FlashLogging
91 Citations2009Shimin Chen
This paper proposes FlashLogging, a logging solution that exploits multiple (USB) flash drives for synchronous logging that achieves up to 5.7X improvements over magnetic-disk-based logging, and obtains up to 98.6% of the ideal performance.
Positional update handling in column stores
75 Citations2010Sándor Héman, Marcin Żukowski +3 more
A new data structure for maintaining such positional updates to columnar databases, called the Positional Delta Tree (PDT), is described, and detailed algorithms for PDT/column merging, updating PDTs, and for using PDTs in transaction management are described.
Proceedings of the VLDB EndowmentOnline maintenance of very large random samples on flash storage
70 Citations2008Suman Nath, Phillip B. Gibbons
Lecture notes in computer scienceCharacterizing the Performance of Flash Memory Storage Devices and Its Impact on Algorithm Design
51 Citations2008Deepak Ajwani, Itay Malinger +2 more
It is shown that despite the similarities between flash memory and RAM (fast random reads) and between flash disk and hard disk (both are block based devices), the algorithms designed in the RAM model or the external memory model do not realize the full potential of the flash memory devices.
Evaluating and repairing write performance on flash devices
46 Citations2009Radu Stoica, Manos Athanassoulis +2 more
This paper proposes new a flash aware data layout --- append and pack --- which stabilizes device performance by eliminating random writes and assesses the impact of append andpack on OLTP workload performance using both an analytical model and micro-benchmarks.
Elsevier eBooksLachesis
29 Citations2003Jiří Schindler, Anastassia Ailamaki +1 more
This paper presents a new storage manager architecture, called Lachesis, that exploits and adapts to observable device-specific characteristics in order to achieve and sustain high performance in database systems.
Lazy-Update B+-Tree for Flash Devices
29 Citations2009Sai Tung On, Haibo Hu +2 more
Simulation results show that the proposed lazy-update method, along with a well-designed commit policy, greatly improves the update performance of the traditional B+-tree while preserving the query efficiency.
Buffered Bloom Filters on Solid State Storage.
28 Citations2010Mustafa Canim, George A. Mihaila +3 more
This paper is proposing a technique to reduce the memory requirement for Bloom filters with the help of solid state storage devices (SSD), and shows that with significantly less memory requirement and fewer hash functions the proposed technique reduces the false positive rate effectively.
Data Science JournalReport from the 6th Workshop on Extremely Large Databases
24 Citations2013Daniel Liwei Wang, Jacek Becla +1 more
This paper is the final report of the discussions and activities at the workshop on extremely large databases, and focuses on practical solutions or influencing DBMS vendors.
PR-join
23 Citations2010Shimin Chen, Phillip B. Gibbons +1 more
This work proposes a new non-blocking join algorithm, Partitioned expanding Ripple Join (PR-Join), which achieves considerably higher early result rates than previous non- blocking joins, while also delivering fast end-to-end query times.