vertica_new_era_in_dbms_performance.pdf

《vertica_new_era_in_dbms_performance.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、TheVerticaAnalyticDatabase–IntroducingaNewErainDBMSPerformanceandEfficiency·Stored20TBofdataon$30,000worthofhardwareandquerieditinunder4minutes·Queried214xfasterthanaproprietarydatawarehousehardwareappliance·Loadeddata3xfasterthanamajorOLTPdatabaseReadaboutthesebe

2、nchmarkresultsandtheinnovativeVerticaAnalyticDatabasearchitecturethatproducedthem.THETIMEFORDATABASEINNOVATIONISNOWAsbusinessesbecomemoreanalyticaltogaincompetitiveadvantageandcomplywithnewregulations,theyarechangingthedemandsplacedondatawarehousesanddrivingdemand

3、fornewquery-intensivedatabaseapplications.Thetablebelowillustrateshowdatawarehouseworkloadshavechangedoverrecentyearsastheyarepushedtoanswermoread-hocquestionsfrommorepeopleanalyzingvastlylargervolumesofdatainnearreal-time.Yet,duringthelastYesterdayToday30years,th

4、erehasDatavolume100sGB100sTBbeenlittledatabaseStandardreports100s10,000smanagementAd-hocqueryusers10s100ssystem(DBMS)GranularityAggregateDetailrecordsinnovationtokeepDataloadingNightlyContinuouslypace.PerformingadhocqueriesonAccessReportsReports,plus…SOAapps,dashb

5、oardswithalertssuchlargedatavolumesdoesnotcomenaturallyforexistingDBMSs,whichusearow-orienteddesignoptimizedforwrite-intensivetransactionprocessingworkloadsratherthanforread-intensiveanalyticalworkloads.Desperateforbetterperformance,row-orientedDBMScustomersspendm

6、illionsofdollarsannuallyonstop-gapmeasuressuchasaddingDBAresources,creatingandmaintainingOLAPcubesorreplacingtheirDBMSwithexpensive,proprietarydatawarehousehardware.LedbydatabasetechnologypioneerDr.MichaelStonebraker,Verticahasdevelopedabreak-throughSQLdatabasetha

7、toffersallsizesofcompaniesacompetitivelyadvantageousandcost-effectiveVerticarequiredjust5newwaytoanalyzeterabytesofinformation.Customerbenchmarksservers,with1TBofshowthattheVerticaAnalyticDatabase:localdiskeachtostore21TBofCDR·Answersqueries50-200xfasteronaveraged

8、ataandprovide·Requires10x-20xlessstoragethanotherdatabasesduetoqueryresultsinlessaggressiveuseofcompressionthan4minuteson·ReducesDBMShardwarecostsbyupto