TAILIEUCHUNG - Self-Tuning Database Systems: A Decade of Progress

The role of the workload, including queries and updates, in physical design was widely recognized. Therefore, at a high level, the problem of physical database design was - for a given workload, find a configuration, . a set of indexes that minimize the cost. However, early approaches did not always agree on what constitutes a workload, or what should be measured as cost for a given query and configuration. Papers on physical design of databases started appearing as early as 1974. Early work such as by Stonebraker [63] assumed a parametric model of the workload and work. | Self-Tuning Database Systems A Decade of Progress Surajit Chaudhuri Vivek Narasayya Microsoft Research Microsoft Research surajitc@ viveknar@ ABSTRACT In this paper we discuss advances in self-tuning database systems over the past decade based on our experience in the AutoAdmin project at Microsoft Research. This paper primarily focuses on the problem of automated physical database design. We also highlight other areas where research on self-tuning database technology has made significant progress. We conclude with our thoughts on opportunities and open issues. 1. HISTORY OF AUTOADMIN PROJECT Our VLDB 1997 paper 26 reported our first technical results from the AutoAdmin project that was started in Microsoft Research in the summer of 1996. The SQL Server product group at that time had taken on the ambitious task of redesigning the SQL Server code for their next release SQL Server . Ease of use and elimination of knobs was a driving force for their design of SQL Server . At the same time in the database research world data analysis and mining techniques had become popular. In starting the AutoAdmin project we hoped to leverage some of the data analysis and mining techniques to automate difficult tuning and administrative tasks for database systems. As our first goal in AutoAdmin we decided to focus on physical database design. This was by no means a new problem but it was still an open problem. Moreover it was clearly a problem that impacted performance tuning. The decision to focus on physical database design was somewhat ad-hoc. Its close relationship to query processing was an implicit driving function as the latter was our area of past work. Thus the paper in VLDB 1997 26 described our first solution to automating physical database design. In this paper we take a look back on the last decade and review some of the work on Self-Tuning Database systems. A complete survey of the field is beyond the scope of this paper. Our .

• Cơ sở dữ liệu
• Technological changes
• Infrastructure development
• Inventory of resources
• Endowment opportunities
• Threats to atmospheric resources
• realizing opportunities
• ATMA Agriculture technological
• Management agency
• Changes in knowledge
• Changes in skill and changes
• Adoption of farm
• natural resources
• alternative measures
• substitution possibilities
• resource scarcity
• Important caveats