Real-time Streaming Insight & Time Series Data Analytic

Transcription

Real-time Streaming Insight & Time SeriesData AnalyticFor Smart RetailSudip MajumderSenior Director – DevelopmentIndustry IoT & Big DataUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/2016

Economic Characteristics of Data“Data is the New Oil” .then “Analytics is the new fuel”Oil ---- FuelData ---- AnalyticsOil is raw and is of little direct useData is raw and is of little direct useOil has Potential EnergyData has Potential ValueGas (refined Oil) has more 5x to 10x the PotentialEnergy than oilAnalytics (refined Data) has more Potential Value thanDataBurning Gas to create motion converts Potential Energyto Kinetic EnergyApplying data sciences to optimize decisions convertsPotential Value to Kinetic ValueOil is converted into fuel that powers the economyData is converted into Analytics that powers thebusinessUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.

4V s Meeting tionContextCognitionUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.Oracle Confidential –2

eBandwidth tohandle datapipeline, embeddedcapacitySort out data , cleansing,Ingest,extract , transform ,loadDescriptive,diagnostics, predictiveand prescriptive modelVelocityNetwork latency,high speed Wi-Fi,Internetconnection, HaLowTimely processing,streaming, near realtime, seamless scale-outJust-in time analytics,on-demandintelligence ,hybridVarietyFlexibility onprotocols e.g. Http,CoAP, MQTT,WebsocketShut down devicesor disconnect thenetwork as neededSheer amount ofdata generated andcollected in sensors,mobile devices,wearable etc.Fastrequest/responsesor one-way trafficbetween devices andbackend systemsDifferent formats ofdata, diversedevices, proprietaryappsShield the noise orbad dataLocation data,clickstream, userbehavior, motionSift thru all data for360 degree view andconduct persuasiveanalyticsSmartrecommendation withengaging relevant andtrustworthy sourcesVeracityFilter dirty/false dataand correlatingapplicable data forsemanticsUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.

Some Bold Imperatives: Without IoT – Companies React to Failure From Sensors to Make Sense Radical Departure From Past to Digital FabricUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.

Complexity of the Consumer Experience Demands Integrated AnalyticsHow well can you anticipate and serve the needs of your consumer at each step of the shopping process?Create DemandFulfill DemandUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Retain and Grow DemandCopyright 2016, Oracle and/or its affiliates. All rights reserved.Copyright 2014 Oracle

Realities of Retail – Inefficiencies and Opportunities No integration of data or analytics across functions.Lack of measurement standards and common metrics.No alignment to the Consumer Experience.Pervasive Silos (Online-In Store Merchandising). No single view of the customer or inventory. Unresponsive supply chain. A Consumer Experience held back by legacy systems.Use or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.Copyright 2014 Oracle

Opportunity: Integrate and Enable New AnalyticsCreate Demand Each functionunderstands its role inserving the consumer. Insights and actionsdirected by a common setof metrics. Discovery andcollaboration becomespossible across multiplebusiness dimensions.Fulfill DemandRetain and Grow Demand IT becomes a strategicenabler.KEY QUESTION:How do you get started?Use or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.Copyright 2014 Oracle

Smart Retail: Process Flow1WEATHERPOS3EventProcessorOracle NoSQLPOSLocalStorageOPSORDM on DBCSINVStructuredSystem ud Service6SMS Alert to CustomerPost Request for SMSEntering Geo FenceProperty GraphSpatial ReportsProperty Graph Segmentas “Node” & Purchase as“Line15Spatial for Item Location in 17Geo MapDatabaseInstanceDerived Tables13OracleIntegrationCloud ServiceUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Customer Purchase14Behavior and Loyalty ,InformationCopyright 2016, Oracle and/or its affiliates. All rights reserved.Call CenterCommunicationwith CustomerJAVA API to LoadDB with baggagedata for DW & Deep 5Learning117CoherenceBig Data42REST API for“Pull” in-memoryand Historical10AnalyticsDashboard,ReportsMetadata inNOSql”Decoding9PublishingWeb Socket 8for “Push”CollectionOracle IoTCloud ServiceSENSORSReal time Capture &Streaming AnalyticsBig Data Discovery16

High Level ArchitectureOracleDB/ExadataCloud ServiceOracle IoTCloud ServiceSENSORSHTTPReceiverSMS AlertWEATHERPOSData LoaderIndustry DataModelPOSDatabaseInstanceLocalStorageDerived TablesOracleIntegrationCloud ServiceReal Time StreamingUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.Confidential – Oracle12

High Level ArchitectureOracleDB/ExadataCloud ServiceOracle IoTCloud ServiceSENSORSHTTPReceiverSMS AlertWEATHERPOSBig DataResponsive & Material DesignIn MemoryOracle No SQLWeb torageREST ServicesPOSBig Data Discovery & MiningReal time Capture &Streaming AnalyticsData Access &PresentationReportingData miningData LoaderIndustry DataModelDatabaseInstanceDerived TablesOracleIntegrationCloud ServiceReal Time StreamingUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.Confidential – Oracle13

Our Solution – Real Time Analytics ArchitectureProduction ViewSENSORSReal Time Capture, StreamingAnalytics &Big Data DiscoveryORDM (Retail DW)INVIn MemoryData StoreJAVA API to LoadDB withException and/orAggr data forDW & DeepLearningLOYALTYCHANNELCRMStructured System of RecordsDatabaseInstanceREST API for “Pull” inmemory and HistoricalSmart Query ProcessorEvent ProcessorData AcquisitionEnrichedData StorePublishingPOSRaw andLong TermData StoreWeb Socket for “Push”AnalyticsPOSHybrid Data StoreDecodingWEATHERDashboard, ReportsMetadata in NOSql”CollectionData could be forked toother processes here.Micro Services Example:SMS Alert to Marketing OpsReal Time UtilizationReportingCustomer PurchaseBehavior, ServiceHistory, CEI InformationCommunicationwith CustomerCustomerExperience DataProperty Graph &Spatial ReportsDerived TablesIn Database AnalyticsUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Property Graph MapCopyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201614

Oracle Solution – Real Time Analytics ArchitectureArchitectural Components Description Feed Orchestrator Hybrid Data Store Smart Query Processor Actionable Intelligence PlatformUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201615

Oracle Solution – Real Time Analytics ArchitectureFeed OrchestratorThe Feed Orchestration Layer provides capture and distribution of records received fromthe data source to downstream processes. Designed to support horizontal scale by adding / removing capacity to meet processing demands Provides Lambda Architecture capability to the solution by enabling writing on multiple types of datastores Provides feeds to: Real-time Message Queue/Stream Processing Modules Relational SQL Database Writers for Quick Response Data Storage NoSQL Database Writers for Intermediate Storage and Processing Hadoop Data Store for Cold Storage and Batch Processing External sources such as Wholesale partnersUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201616

Oracle Solution – Real Time Analytics ArchitectureHybrid Data StoreThe Hybrid Data Store consists of: Hadoop Data Store NoSQL/Real Time Data Store Relational Data StoreUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201617

Oracle Solution – Real Time Analytics ArchitectureHybrid Data StoreThe Hybrid Data Store provides Peta scale capability to handle: Real Time / Near Real Time Streams SQL Data store for Relational data store and self-serve reporting data model NoSQL Data store for Schema less data store and transaction level events processing Hadoop Data store for Long Term Storage of Raw data / transactions Provides flexible linear horizontal and vertical scaling capability - expansion of individualdata stores Serves both the Batch Layer and Speed Layer in support of a Lambda ArchitectureUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201618

Oracle Solution – Real Time Analytics ArchitectureSmart Query ProcessorThe Smart Query Layer provides the capability to query multi-platform back end datastores using standards-based SQL tools Provides singular interface for querying the Hybrid Data Store Enables retrieval and merging of query results from Near Real Time feeds and SQL/NoSQL storage Provides JDBC Connectivity to downstream applications Provides load balancing capabilities to query servers Acts as the Serving Layer of the Lambda ArchitectureUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201619

Oracle Solution – Real Time Analytics ArchitectureActionable Intelligence PlatformThe Actionable Intelligence Platform is composed of 3 modules Real Time Insights Based on open source / Angular JS Provides capability to query and report on real time feeds, create dynamic dashboards and canned reports Provides flexibility to internal development teams to create rich visualization for specific users or use cases OBIEE / VA (Optional) Provides Self-Service, Dash boarding and Reporting capabilities Provides pre-built reports and dashboards for use by various Stake Holders, such as Store Ops, Marketing, Finance, CategoryManagement, Loyalty Program, etc. Big Data Discovery (Optional) Provides the ability to explore, hypothesize and validate the scenarios for critical business problems and potential businessopportunities Provides test bed interface to analyze data without impacting production reportsUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.10/5/201620

WorkflowHi speed data extractionAnalytics and ReportingEnginePresentationPredictions and decisionscollectorAnalysis engine Logs are collected from POSMachine, RFID tags, IOT devices etc.with utmost efficiency not to hamperthe device performance. As the data capture size if huge, itneeds significant amount of storage.There comes filtering that helps youextract selected data rather than alland save you from running out ofstorage space. Streaming analysis is muchfaster and near real time. Streaming analytics based onEsper CEP engine to generatecontinuous KPI stream Data analytics and decisionengine, persistence storageand data mining capabilitiesbuilt using Apache Hadoopcomponents for horizontalexpandability and faster queryexecution Data modeling andreporting built using Hiveand R. Visualization of outcomethrough multi-channelpresentation layerrelevant to individualstakeholders fromdashboard view to drilldown capability ofcorrelated transactiontracking Responsive UI designUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response. Identify business prospectsForesee challengesLogging and surveillanceGenerate innovative business plansand pricing schemeCopyright 2016, Oracle and/or its affiliates. All rights reserved.

Key Take-a-Ways Like the Internet, IoT Monetization will occurValue delivery will dictate early winnersIoT Monetization silver bullet is packaging and pricing flexibilityReal time visibility and agility are essential to monetizationScale goes beyond volume, infrastructure needs to support business scaleUse or disclosure of the data contained on this sheet is subject to the restrictions on the front page of this response.Copyright 2016, Oracle and/or its affiliates. All rights reserved.

ORDM (Retail DW) Experience Data Database Instance Derived Tables In Database Analytics llection g g cs Real Time Capture, Streaming Analytics & Big Data Discovery Real Time Utilization Reporting REST API for “Pull” in-n ocessor Raw and memory and Historical Long Term Data Store or Property Graph & Spatial Reports Hybrid Data Store Data could be forked to