{"title":"基于闪存SSD的DBMS,用于高性能计算嵌入式和多媒体系统","authors":"S. S. Rizvi, Tae-Sun Chung","doi":"10.1109/ICCES.2010.5674850","DOIUrl":null,"url":null,"abstract":"Flash memory based large capacity SSDs open the doors on high performance computing applications by offering remarkable throughput and amazing reliability. However, flash memory hardware characteristics like erase-before-write and limited endurance cycles do not allow disk based schemes implication directly. For employing such schemes, we need to revise them on some level to make them effective for flash media. Regarding, the researchers have devoted considerable efforts to implement the DBMSs on flash for fast retrieval of data that is bit indolent using file systems. However, previous techniques require huge main memory space and high computational power for data processing. This paper proposes an advanced DBMS architecture using key sequenced data set, virtual sequential access method and multilevel indexing for flash memory SSD based performance oriented embedded and multimedia applications. Basic database operations plus space reclamation and memory wear-leveling are achieved by taking flash characteristics into account carefully. Main memory based buffer management is implemented to increase throughput and for efficient media utilization. Comprehensive performance evaluations using two modern benchmarks prove less overhead with high reliability and outstanding throughput for flash SSD based DBMSs compare to HDD.","PeriodicalId":124411,"journal":{"name":"The 2010 International Conference on Computer Engineering & Systems","volume":"42 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2010-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"Flash memory SSD based DBMS for high performance computing embedded and multimedia systems\",\"authors\":\"S. S. Rizvi, Tae-Sun Chung\",\"doi\":\"10.1109/ICCES.2010.5674850\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Flash memory based large capacity SSDs open the doors on high performance computing applications by offering remarkable throughput and amazing reliability. However, flash memory hardware characteristics like erase-before-write and limited endurance cycles do not allow disk based schemes implication directly. For employing such schemes, we need to revise them on some level to make them effective for flash media. Regarding, the researchers have devoted considerable efforts to implement the DBMSs on flash for fast retrieval of data that is bit indolent using file systems. However, previous techniques require huge main memory space and high computational power for data processing. This paper proposes an advanced DBMS architecture using key sequenced data set, virtual sequential access method and multilevel indexing for flash memory SSD based performance oriented embedded and multimedia applications. Basic database operations plus space reclamation and memory wear-leveling are achieved by taking flash characteristics into account carefully. Main memory based buffer management is implemented to increase throughput and for efficient media utilization. Comprehensive performance evaluations using two modern benchmarks prove less overhead with high reliability and outstanding throughput for flash SSD based DBMSs compare to HDD.\",\"PeriodicalId\":124411,\"journal\":{\"name\":\"The 2010 International Conference on Computer Engineering & Systems\",\"volume\":\"42 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2010-12-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"The 2010 International Conference on Computer Engineering & Systems\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICCES.2010.5674850\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"The 2010 International Conference on Computer Engineering & Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICCES.2010.5674850","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Flash memory SSD based DBMS for high performance computing embedded and multimedia systems
Flash memory based large capacity SSDs open the doors on high performance computing applications by offering remarkable throughput and amazing reliability. However, flash memory hardware characteristics like erase-before-write and limited endurance cycles do not allow disk based schemes implication directly. For employing such schemes, we need to revise them on some level to make them effective for flash media. Regarding, the researchers have devoted considerable efforts to implement the DBMSs on flash for fast retrieval of data that is bit indolent using file systems. However, previous techniques require huge main memory space and high computational power for data processing. This paper proposes an advanced DBMS architecture using key sequenced data set, virtual sequential access method and multilevel indexing for flash memory SSD based performance oriented embedded and multimedia applications. Basic database operations plus space reclamation and memory wear-leveling are achieved by taking flash characteristics into account carefully. Main memory based buffer management is implemented to increase throughput and for efficient media utilization. Comprehensive performance evaluations using two modern benchmarks prove less overhead with high reliability and outstanding throughput for flash SSD based DBMSs compare to HDD.