資深總監(jiān)*前述特定亞馬遜云科技生成式人工智能相關(guān)的服三個哲學(xué)問題*前述特定亞馬遜云科技生成式人工智能相關(guān)的服??2022.06,650+Employees?Across9globaloffices?2020.11,$270MRoundD?2020.01,TiDBhits1000usersglobally?2015.04,Startingpoint*前述特定亞馬遜云科技生成式人工智能相關(guān)的服lModernlModernSQLdatabaselFull-featuredSQLlMySQLcompatibilitylACIDcompliancelHighAvailabilitywithstrongconsistencylServebothOLTP&Real-timeOLAPScalabilityiseverything*前述特定亞馬遜云科技生成式人工智能相關(guān)的服3,000+globaladoptersuseTiDBinproduction...China/Global.theWorld.And2outoftop3CryptocurrencyExchanges*前述特定亞馬遜云科技生成式人工智能相關(guān)的服*前述特定亞馬遜云科技生成式人工智能相關(guān)的服Scalabilityiseverything2010第—次閃崩:2010第—次閃崩:400年—遇2025四月閃崩:三次TheDJIAonMay6,2010(11:00AM–4:00PMEDT)DOW:998.5points(~9%)*前述特定亞馬遜云科技生成式人工智能相關(guān)的服TSO/DataTSO/DatalocationMetadataApplicationviaApplicationvia*前述特定亞馬遜云科技生成式人工智能相關(guān)的服TSO/DataTSO/DatalocationMetadataApplicationviaApplicationvia*前述特定亞馬遜云科技生成式人工智能相關(guān)的服TSO/DatalocationMetadata/ScTSO/DatalocationApplicationviaLightningAnalyticalEngineApplicationviaLightningAnalyticalEngineTiDBOperatorAndotherprovidersPublicAndotherprovidersPublicCloud*前述特定亞馬遜云科技生成式人工智能相關(guān)的服站在前人的肩膀上–1978/2012Time,Clocks,andtheOrderingofEventsinaDistributedSystem,Communicationof輯時鐘”概念，2013圖靈獎獲得者Spanner:Google‘sGlobally-DistributedDatabase,ProceedingsofOSDI，2012---強一致分布式事務(wù)系統(tǒng)實現(xiàn)，工業(yè)級NewSQLFirstdistributedHTAP-2020TiDB:ARaft-basedHTAPDatabase,UniKV:TowardHigh-PerformanceandScalableKVStorageiMatrixKV:ReducingWriteStallsandWriteAmplificationinLSM-treeBasedKVStoreswithMatrixContainerinNVM,USEINBuildingAFastandEfficientLSM-treeStorebyIntegratingLocalStoragewithCloudStorage,ACM，TransactionsonArchitectureandCodeOptimization2022TriangleKV:ReducingWriteStallsandWriteAmplificationinLSM-treDifferentiatedKey-ValueStorageManagementforBalancedI/OPerformance,USEINX2021BuildingAFastandEfficientLSM-treeStorebyIntegratingLocalStoragewithCloudStoComboTree:APersistentIndexingStructurewithUniversalOperationalEfficiencyandScalability,IEEETPDS2021MakeitSmarter-2025Hyper:HybridPhysicalDesignAdvisorwithMulti-agentReinforcementLearning,ICDE2025MachineLearningInferencePipelineExecutionUsingPureSQLBasedonOperatorFusion,ICDE2025Chat2query:Azero-shotautomaticexploratorydataanalysissystemwithlargelanguagemodels,IEEE2024Autotqa:Towardsautonomoustabularquestionansweringthroughmulti-agentlargelanguagemodels,VLDB2024OSSInsight:ScalableGitHubAnalysis,VLDB2024*前述特定亞馬遜云科技生成式人工智能相關(guān)的服務(wù)僅在亞馬遜云科技海外區(qū)域可用,亞馬遜云科技中國僅為幫助您發(fā)展海外業(yè)務(wù)和/或了解行業(yè)前沿技術(shù)選擇推薦該服務(wù)。*前述特定亞馬遜云科技生成式人工智能相關(guān)的服務(wù)僅在亞馬遜云科技海外區(qū)域可用,亞馬遜云科技中國僅為幫助您發(fā)展海外業(yè)務(wù)和/或了解行業(yè)前沿技術(shù)選擇推薦該服務(wù)。Journeystartin2019*前述特定亞馬遜云科技生成式人工智能相關(guān)的服TiDBSelf-ManagedApplicationInfraApplicationInfra*前述特定亞馬遜云科技生成式人工智能相關(guān)的服*前述特定亞馬遜云科技生成式人工智能相關(guān)的服TiDBSelf-ManagedTiDBCloudDedicated350TB2024.082024.09,migratetocloud2024.12,scaleupxNthanwedidsmall-scale,self-manage*前述特定亞馬遜云科技生成式人工智能相關(guān)的服3queriesAmazonQuickSight24AmazonAmazon15AmazonGlueApplicationAmazonTiDBCloudcaningesttransactionaldatafromapplicationsorstreamingdatafrom3queriesAmazonQuickSight24AmazonAmazon15AmazonGlueApplicationAmazonTiDBCloudcaningesttransactionaldatafromapplicationsorstreamingdatafromWhileservingOLTPworkloads,TicanautomaticallyconvertrowtocolumnarUsingTiDBCloudHTAPfeatuanalyticscanbeperformedusingQuickSight.downstreamprocessing.Oncereplicated,thedatacanbeenriched,aggregated,andanalyzedwithadatalakeanddatawarehousearchitectureusingAmazonS3,AmazonL12345*前述特定亞馬遜云科技生成式人工智能相關(guān)的服ScalabilityIssues:Traditionaldatabasesystemsstruggledtohandlethegrowingdemandsofonlinegames,ScalabilityIssues:Traditionaldatabasesystemsstruggledtohandlethegrowingdemandsofonlinegames,especiallyduringpeaktraffic.PerformanceBottlenecks:Latencyandperformanceissuesimpactedreal-timedataprocessing,affectinguserexperience.HighMaintenance:Significantresourceswerespentondatabasemanagement,increasingoperationalcosts.Scalability:TiDBCloudenableshorizontalscalingwithoutdowntime,ensuringsmoothperformanceevenduringtrafficspikes.Real-TimeDataProcessing:SupportsbothOLTPandOLAPworkloads,reducinglatencyandimprovinguserexperience.ReducedOverhead:Fullymanagedservicelowersoperationalcomplexityandcosts.ProvenPerformance:TiDBCloudcanefficientlyhandlelargevolumesoftraffic,supportingthesuccessofCapcom’sglobalonlinegames.knownforhitslikeMonsterHunter,whichhassold84millionworldwide(asofJuly2022),andthepopularforover25years.Industry:Gaming*前述特定亞馬遜云科技生成式人工智能相關(guān)的服Distributed-NativeCloud-NativeAI-NativeManus:鏈?zhǔn)絽f(xié)作式AgentAI門檻降低→中小創(chuàng)業(yè)公司入局,Multi-Modal多模,SaaS革命時代*前述特定亞馬遜云科技生成式人工智能相關(guān)的服lUnlimitedScalabilitylUnlimitedScalabilitylMulti-WriteslAuto-Split-StoragelStorageEfficiencylMicroServicesforbackup/compaction/log/Co-*前述特定亞馬遜云科技生成式人工智能相關(guān)的服Max:sum(all_max_value)/(allAvg:sum(all_avg_value)/(allCPU)=20%*前述特定亞馬遜云科技生成式人工智能相關(guān)的服AppAppAmazonObjectStorageObjectStorage*前述特定亞馬遜云科技生成式人工智能相關(guān)的服ReduceTCO(TotalCostofOwnership)upto20~70%*DedicatedModeNext-GenMode1x%~5x%5x%~7x%Total5x%~6x%*前述特定亞馬遜云科技生成式人工智能相關(guān)的服User-AImodelconversationUser-AImodelconversationinteractionsystemunderthe?Atablewithovertenbillionrows?Neededadistributeddatabasetomeettherapiddevelopmentofthe?Scalability:Automatedscaling,includestorageandcompute.?ApplicationFriendly:UseTiDBassingle-nodedatabase,noinvasiontoapplication?HA:Financial-gradehighavailabilitycapabilitiesP99<50ms-ExcellentPerPBcapacity-Unlimitedstoragenode*前述特定亞馬遜云科技生成式人工智能相關(guān)的服?Atablehadover100millionrows,andthecurrentproduct'sperformanceishittingabottleneck?Neededtocompletethetestingandgolivewithin3weeks?Scalability:Automatedscaling,includestorageandcomputeGolivein3weeks-MySQLcompatiblitySecond-levelsynchronizationbetweendifferentclusters-TiCDCRecordandquerythestatusandinformationoftheAIagentateachstage.*前述特定亞馬遜云科技生成式人工智能相關(guān)的服?B