TAILIEUCHUNG - Lecture Database system concepts - Chapter 21: Parallel databases

This chapter is suitable for an advanced course, but can also be used for independent studyprojectsbystudentsofa first course. The chapter covers several aspects of the design of parallel database systems partitioning of data, parallelization of individual relational operations, and parallelization of relational expressions. The chapter also briefly covers some systems issues, such as cache coherency and failure resiliency. | Chapter 21: Parallel Databases Database System Concepts, 5th Ed. ©Silberschatz, Korth and Sudarshan See ­ for conditions on re­use Chapter 21: Parallel Databases s Introduction s I/O Parallelism s Interquery Parallelism s Intraquery Parallelism s Intraoperation Parallelism s Interoperation Parallelism s Design of Parallel Systems Database System Concepts ­ 5th Edition, Aug 22, 2005. 21. ©Silberschatz, Korth and Sudarshan Introduction s Parallel machines are becoming quite common and affordable q q Prices of microprocessors, memory and disks have dropped sharply Recent desktop computers feature multiple processors and this trend is projected to accelerate s Databases are growing increasingly large q q large volumes of transaction data are collected and stored for later analysis. multimedia objects like images are increasingly stored in databases s Large­scale parallel database systems increasingly used for: q storing large volumes of data q processing time­consuming decision­support queries q providing high throughput for transaction processing Database System Concepts ­ 5th Edition, Aug 22, 2005. 21. ©Silberschatz, Korth and Sudarshan Parallelism in Databases s Data can be partitioned across multiple disks for parallel I/O. s Individual relational operations (., sort, join, aggregation) can be executed in parallel q data can be partitioned and each processor can work independently on its own partition. s Queries are expressed in high level language (SQL, translated to relational algebra) q makes parallelization easier. s Different queries can be run in parallel with each other. Concurrency control takes care of conflicts. s Thus, databases naturally lend themselves to parallelism. Database System Concepts ­ 5th Edition, Aug 22, 2005. 21. ©Silberschatz, Korth and Sudarshan I/O Parallelism s Reduce the time required to retrieve relations from disk by partitioning s the relations on multiple disks. s .

• Cơ sở dữ liệu
• Database system concepts
• Lecture Database system concepts
• Database systems
• Database languages
• Database design
• Parallel databases
• Database system architectures
• Relational database design
• Recovery system
• Query optimization
• Transaction state
• Concurrency control
• Data analysis
• Information retrieval
• Distributed databases
• Advanced application development
• Advanced data types
• Advanced transaction processing
• Relational databases
• Relational model
• Basic query structure
• Dynamic SQL
• Tuple relational calculus
• Entity relationship model
• Application design
• Object­ based databases
• XML applications
• Physical storage media
• Static hashing
• Query processing