從 DIY 轉向 AI 準備.pdf

《從 DIY 轉向 AI 準備.pdf》由會員分享，可在線閱讀，更多相關《從 DIY 轉向 AI 準備.pdf（24頁珍藏版）》請在三個皮匠報告上搜索。

1、Move from DIY Move from DIY to ready for AIto ready for AIEvolving your data architecture for scalable successMove from DIY to Move from DIY to ready for AIready for AIFIVETRANFIVETRANThe average organization aggregates data from over 400 sources.While traditional DIY pipelines may offer benefits li

2、ke control and customization,they come at the expense of scalability,reliability and maintenance costs.In our talk,you will discover:The complexity of scaling data movement with DIY data pipelines and the challenges it presents How to better control costs and streamline operations using a managed,au

3、tomated data pipeline platform The benefits of leveraging a modern,interoperable platform to enhance performance for advanced analytics and AI initiativesChris RudolphChris RudolphLead Sales Engineer,Enterprise2In the expectations of data teams in the industry and by the business3In the complexity o

4、f the problems for data teams to solve for the business4Data practitioners will shape how GenAI is deployed in the enterprise.Source:MIT Report,2024Proportion of data scientist TIME SPENT TIME SPENT PREPARING DATAPREPARING DATA on averageHow time is spent on AI projectsHow time is spent on AI projec

5、ts567%33%Proportion of data scientist TIME SPENT TIME SPENT BUILDING AI MODELSBUILDING AI MODELS on averageSource:Vanson Bourne,2024How often data pipelines have to be rebuilt How often data pipelines have to be rebuilt after being deployedafter being deployed641%41%of data leaders report that they

6、rebuild data pipelines sometimes.18%18%of data leaders report that they rarely rebuild data pipelines39%39%of data leaders report that they constantly rebuild data pipelines2%2%of data leaders report that they never rebuild data pipelinesSource:Wakefield Research,20217The DIY data pipeline icebergTh

7、e DIY data pipeline icebergMove data from source to destinationMove data from source to destinationCopying data is easy!Data modelingData modelingYou all know SQL,right?Schema drift handling Schema drift handling I cant believe the pipeline broke because the app added a new field.AutomationAutomatio

8、nI thought this was supposed to run on a schedule.What happened?IdempotenceIdempotenceWhy do we have so many duplicate records?Scaling and extensibility Scaling and extensibility We keep hiring engineers,yet things dont seem to get better Security and PID management Security and PID management This

9、is bad.were$4M in the hole because our pipeline leakedMythbustingMythbustingDiving into the factsFALSEFALSEDIY data pipelines are DIY data pipelines are more secure.more secure.TRUETRUEFALSEFALSE Decreased security capabilities.Slow adaptation to schema changes.Lack of an incident response plan.Lack

10、 of proper error handling.Security&governanceSecurity&governance10Modern Data Integration tools have local deployment options where data never leaves your environmentFALSEFALSEDIY data pipelines offer DIY data pipelines offer customization.customization.TRUETRUEFALSEFALSECustomizationCustomization T

11、oo much customization can impede interoperability.Modern tools offer the right degree of customization.Modern tools offer scalability.12FALSEFALSEDIY data pipelines are DIY data pipelines are cheaper.cheaper.TRUETRUEFALSEFALSECostsCosts Various high hidden costs.Longer time-to-value.Modern integrati

12、on tools can help do more with less.Data engineering teams can gain productivity by 48%.Faster time to insights result in operational efficiencies.Increased focus on strategic initiatives as data teams spend less time to deliver insights.1466%66%33%33%more effective at replicating data for analytics

13、less data team time to deliver insights15“Fivetran has revolutionized our approach to data,enabling AI/ML and GenAI initiatives on employee performance and providing managers with faster insights.These advancements were unimaginable just two years ago.”Sandro FratturaAnalytics Engineering Manager,Hu

14、bSpot People OperationsDATA STACKFivetran customers innovate with reliablereliable,scalablescalable and securesecure data integration to power real-time analytics,operational efficiency and artificial intelligenceartificial intelligence.A complete enterpriseA complete enterprise-grade platformgrade

15、platform18I OTReal-t i me per sonal i zat i onSuppl y Chai n M anagementet c Data SourcesDataintegration platformTarget destinationsData transformationOutputsDatabasesData WarehouseFilesMarketing Anal.Finance Anal.Product Anal.Sales Anal.Event CollectorsReverse ETLBusiness IntelligenceEmbedded Analy

16、ticsAd Hoc ReportingML/AIYour own productExtensibilityObservationOrchestrationGovernanceConnector Connector SDKSDKBuild it your way,automate it with Fivetran.19NEW!NEW!Your own customizable connector:Your own customizable connector:Build a connector on your timeline for any source,move data using yo

17、ur own proprietary APIs,augment our native connectors,or extract specific,tailored data sets from any source.In an easyIn an easy-toto-setup way:setup way:The SDK allows you to code a custom data connector using Python,and a simple Command Line Interface(CLI)command deploys your new connector to Fiv

18、etrans secure cloud environment.The benefits of a managed service:The benefits of a managed service:Leverage the Fivetran platform to handle data-writing,retries and inferring the schema.Rely on our robust failure recovery and security and compliance features to build quickly and get to production f

19、aster.20Why Fivetran Why Fivetran ReliabilityReliabilityReliable performance to ensure your teams valuable time is no longer spent on maintenance.End-to-end automationautomationincluding 99.9%uptime99.9%uptimeacross 3M+daily syncs500+500+team of engineers&customer support ScalabilityScalabilityGrow

20、your team,data volumes and business with a platform that supports all your data sources and destinations.500+500+pre-built connectors Real timeReal time data movementHigh performance,low impact CDCCDCPlatform extensibilityextensibilitySecuritySecurityRigorous,built-in platform security and governanc

21、e to move data with peace of mind.Multiple secure deploymentdeploymentoptionsData governancegovernance,observability&integrityThe most securitymost securitycredentials21$1.5m$1.5mAverage annual benefits*$177.4k$177.4kAnnual operational cost savings$89.5k$89.5kAverage annual benefits per data source$

22、83k$83kAdditional net revenue“My organization is more resilient to business changes.Because of Fivetran,were able to provide access to able to provide access to data a lot quickerdata a lot quickerthan we would have in the past.”RETAIL ORGANIZATIONDrive millions in financial impact and enable Drive

23、millions in financial impact and enable new business initiativesnew business initiativesImprove business operations and increase revenue opportunitiesIncrease the productivity of data engineers,analytics teams,and governance teams by automating highly manual tasksDecrease data silos across teamsRedu

24、ce technical debt and be more operationally resilient to business challengesThe total economic The total economic impact of Fivetran impact of Fivetran“Just because you can write custom code to do data integration or create your own data warehouse built on Hadoop doesnt mean you should,especially for business-critical analytics.Mark KidwellChief Data ArchitectData Platforms and Services,at Autodesk.Thank you!Thank you!Hosted by:Thank you for attending!We value your feedback!Please rate your speaker in the conference evaluation provided at the end of the event.