在线查询处理讲义.pptVIP

Online Query ProcessingA Tutorial Peter J. Haas IBM Almaden Research Center Joseph M. Hellerstein UC Berkeley Goals for Today Exposure to online query processing algorithms and fundamentals Usage examples Basic sampling techniques and estimators Preferential data delivery Online join algorithms Relation to OLAP, etc. Some thoughts on research directions More resources to appear on the web Annotated bibliography Extended slide set Survey paper Road Map Background and motivation Human-computer interaction Tech trends and prognostications Goals for online processing Examples of online techniques Underlying technology Related work Looking forward Human-Computer Interaction Iterative querying with progressive refinement Real-time interaction (impatience!) Spreadsheets, WYSIWYG editors Modern statistics packages Netscape STOP button Visually-oriented interface Approximate results are usually OK Disk Appetite Greg Papadopoulos, CTO Sun: Moores Law Aint Good Enough (Hot Chips ’98) The Latest Commercial Technology Drawbacks of Current Technology Only exact answers are available A losing proposition as data volume grows Hardware improvements not sufficient Interactive systems fail on massive data E.g., spreadsheet programs (64Krow limit) DBMS not interactive No user feedback or control (“back to the 60’s”) Long processing times Fundamental mismatch with preferred modes of HCI OLAP: a partial solution Can’t handle ad hoc queries or data sets Goals for Online Processing New “greedy” performance regime Maximize 1st derivative of the “mirth index” Mirth defined on-the-fly Therefore need FEEDBACK and CONTROL Road Map Background and Motivation Examples of Online Techniques Aggregation, visualization, cleaning/browsing Underlying technology Related work Looking Forward Online Aggregation SELECT AVG(temp) FROM t GROUP BY site 330K rows in table the exact answer: Online Aggregation, cont’d A simple online aggregation interface (after 74 rows) Online Aggregation, cont’d