inform innovate accelerate optimize How telcos can benefit from streaming big data analytics #streamingbigdataanalytics Sponored by: 2013 TM Forum 1 V2013.4
Today s Speakers Adrian Pasciuta Director of Industry Solutions Rebecca Sendel Senior Director, Business Assurance Program 2013 TM Forum 2 V2013.4
inform innovate accelerate optimize Data Analytics Best Practices TM Forum Webinar September 2014 2014 TM Forum 3 V2013.5
Big Data Analytics Identification, design and deployment of strategies, processes, skills, systems and data that can provide actionable intelligence resulting in business value Volume VALUE Velocity Variety 2014 TM Forum 4 V2013.5
Biggest Challenges for Success in Data Analytics 2014 TM Forum 5 V2013.5
Planned/Expected Investment Areas for Analytics 2014 TM Forum 6 V2013.5
Data Analytics in TM Forum Common Language 2014 TM Forum 7 V2013.5
Big Data Analytics Guidebook 2014 TM Forum 8 V2013.5
Data Repository Structured Data, Unstructured Data, Semi-structured Data GB979: Big Data Analytics Reference Model Data Governance Privacy, Security, and Compliance CAPEX Reduction Applications OPEX Reduction Applications CEM Applications Revenue Generating Applications Other Applications Batch Streaming Data Analysis Data Analysis Data Data Modeling, Complex Metrics, Event Reports Processing, Alerts & Triggers, Reports Data Management Transformation, Correlation, Enrichment, Manipulation, Retention Data Ingestion Integration, Import, Format V2013.5 Data Sources Network, OSS, BSS, Social Networks, 2014 TM Forum 9
Big Data Analytics Use Cases Use Case ID Use Case Use Case ID Use Case S-MOM-T1 Real-time Personalized Offers while Browsing O-CRM-PC7 Roaming Customer Onboarding S-MOM-T2 S-MOM-T3 S-MOM-T4 S-MOM-T5 S-MOM-T6 Real-time Personalized Offers during Checkout Real-time Personalized Offers during a Live Interaction Real-time Personalized Offers Based on Location Real-time Personalized Offers Based on Usage Real-time Personalized Offers Based on Device O-CRM-CR1 O-CRM-CR2 O-CRM-CR3 O-CRM-CR4 Churn Risk Prediction for Customer Retention Churn Motivation Prediction for Customer Retention Personalized Offers for Customer Retention Retention Offer Acceptance Propensity Analytics S-MOM-T7 S-MOM-T8 S-MOM-T9 S-MOM-O1 S-MOM-O2 S-MOM-O3 S-MOM-O4 S-SDM1 S-SDM2 S-RDM1 S-RDM2 S-RDM3 O-CRM-CC1 O-CRM-CC2 Intelligent Advertising Based on Browsing History Stimulating Prepaid to Postpaid Conversion Enticing Usage from Roaming Customers Product Definition and Development Product Introduction Analytics Product Performance Optimization Purchase Propensity Analytics for Enhanced Targeting CSP Data Monetization MVNO Data Monetization Value-based Network Planning New Enterprise Order Impact Analysis Policy-based Capacity Management Personalization of Real-Time Interaction in Assisted Care Increase Effectiveness of Customer Self Service O-RMO1 O-RMO2 O-RMO3 O-RMO4 O-SPRM1 O-FUL-O1 O-FUL-O2 O-FUL-I1 O-FUL-I2 O-FUL-I3 O-BRM1 O-BRM2 E-SEP1 Network Fault Location and Recovery Real-time Value-based Congestion Management Real-Time Customer Offload Management Proactive Experience Driven Network Repair Partner Value Optimization Increasing Conversion in the Ordering Process Reduction of Errors in the Ordering Process Optimization of Customer Self-Installation Field Technician Assignment Optimization Field Technician Arrival Optimization Revenue Assurance Personalized Collections Treatment Plan Market Watch O-CRM-CC3 O-CRM-PC1 O-CRM-PC2 Improving Assisted Care with Network Experience Analytics Proactive Care Right Proactive Care Channel and Time E-EEM1 E-FAM1 Business Process Optimization Fraud Management O-CRM-PC3 Proactive Care Based on Poor Care Experience O-CRM-PC4 Proactive Care During or After Network Fault O-CRM-PC5 Proactive Care Based on Absence of Usage O-CRM-PC6 Proactive Care Based on Network Experience Analytics 2014 TM Forum 10 V2013.5
500+ Pre-defined Metrics www.tmforum.org Standards Menu Tools Frameworx Metrics Repository 2014 TM Forum 11 V2013.5
inform innovate accelerate optimize Thank You! 2014 TM Forum 12 V2013.5
Streaming Analytics for Big Data: What s In It for CSPs? Adrian Pasciuta Director, Industry Solutions 17 th September 2014 2014. VITRIA TECHNOLOGY, INC. All rights reserved.
Topics The TMF Big Data Reference Model What is Streaming Analytics? From Streaming Analytics to Operational Intelligence Streaming Analytics in the Big Data Ecosystem What Problems Does It Solve? Customer Use Cases Summary and Q&A 2014 www.vitria.com 14
GB979: TMF Big Data Analytics Reference Model Data Repository Structured Data, Unstructured Data, Semi-structured Data Data Governance Privacy, Security, and Compliance CAPEX Reduction Applications OPEX Reduction Applications CEM Applications Revenue Generating Applications Other Applications Batch Streaming Data Analysis Data Analysis Data Data Modeling, Metrics, Complex Reports Event Processing, Alerts & Triggers, Reports Data Management Transformation, Correlation, Enrichment, Manipulation, Retention Data Ingestion Integration, Import, Format Data Sources Network, OSS, BSS, Social Networks, 2014 www.vitria.com 15
Batch vs. Streaming Analytics On-Demand Streaming In memory Request-based (when you ask) One-time evaluation Bulk algorithms Disk-based (usually) Event-based (when something happens) Continuous evaluation Incremental algorithms In-memory (by design) Investigative Retrospective Retrospective Investigative Reactive Reactive Prospective Predictive Proactive 2014 www.vitria.com 16
Streaming Analytics Required capabilities Streaming Analytics Correlate and Enrich - across diverse sources Correlate across Time - Track and trace Detect Patterns & Trends Advanced Real-time Analytics Predictive Analytics 2014 www.vitria.com 17
Vitria Operational Intelligence Powered by Streaming Analytics Real-Time VISIBILITY Streaming Analytics Immediate ACTION Real-time Information in context Rich Dashboards User Empowerment Correlate and Enrich - across diverse sources Correlate across Time - Track and trace Detect Patterns & Trends Advanced Real-time Analytics Predictive Analytics Respond quickly using Automated Processes & Guided Workflows Location and Situation Awareness 2014 www.vitria.com 18
Collaborative Development Tools Developer and Business Analyst Shared Services & REST APIs Admin Tools & Utility Apps Vitria Operational Intelligence Architecture Visualization Layer Action Layer In Memory Streaming Analytics Input Stream(s) Continuous Queries Output Stream(s) Feed Layer Data Integration Layer Event Sources / Processing / Event Targets ESB Networks / Sensors / Devices Web Services/ Messaging Systems & Applications Databases & Files Social / Weather / Traffic 2014 www.vitria.com 19
Streaming Analytics in Action: A Network Event Processing Pipeline Network Signalling Events Example 250,000+ EPS EPS = Events Per Second Events of Interest ~2,500 EPS Significant Events ~250 EPS Network & IoT Data - Examples Network signalling CDRs Telemetry / smart meter data feeds Streaming data from weather / traffic sources Customer, Network & Device Reference Data Event Filtration Enrich & Store Stage 1: Filter & Enrich Event Filtering Duplicate removals Event enrichment Raw event statistics Storage of events enriched w/ context for look back/hypothesis analysis Probable Match Stage 2: Use Cases Complex Patterns and Rules KPI Computation SLA measurement Frequencies & scoring Multidimensional Analytics Prediction or Trigger Stage 3: Action Confirmed trend to predictive pattern Process driven actions Updates and Alerts Guided workflows Automated multi-step actions Visualizations Stored calculations for analysis 2014 www.vitria.com 20
Elastic Scalability Detect anomalous events Correlate Customer data Multidimensional analysis Scale out on commodity hardware an elastic grid of compute servers Analytic Server Analytic Server Analytic Server ibpm Each use case defined as a multi-step Event Processing Network (EPN) Network events Analytic Server Analytic Server Analytic Server ibpm Each EPN contains multiple Projects pipelined together A Project uses Map-Reduce to scale Analytic Server Analytic Server Analytic Server ibpm Contextual (business) data is preloaded into memory Intelligent BPM enables immediate action Asset & Customer Context Automated Actions, Alerts, Workflows & visualizations 2014 www.vitria.com 21
Predictive Analytics & Intelligent Action Automated Triggers Matched patterns initiate intelligent alerts Intelligent Actions Automated processes Guided human workflows Advanced Analytics Multidimensional, duration, outcome Predictive Analytics & Trending Prescriptive Analytics Next best action Intelligent Processes Adaptive process behavior Based on situational and social awareness, location awareness 2014 www.vitria.com 22
Streaming Analytics in the Big Data Ecosystem Big Data In Motion & At Rest 2014 www.vitria.com 23
Streaming Analytics in the Big Data Ecosystem Big Data In Motion & At Rest Event capture Queries Data & Result Streaming 2014 www.vitria.com 24
Streaming Analytics in the Big Data Ecosystem Complementary Big Data In Motion & At Rest Batch Streaming Event capture Queries Data & Result Streaming Vitria OI + Hadoop Lambda Architecture Raw events are sent to both Vitria OI and Hadoop Vitria OI for streaming analytics (Speed Layer) Hadoop provides historical storage and historical analytics (Batch layer) Vitria OI can also query Hadoop to provide real-time insight with historical context 2014 www.vitria.com 25
What Problems Can I Solve with Streaming Analytics? Real-time Network Optimisation Real-time performance of all network cells Real time dropped call detection Adjacent cell performance (error perturbation) Cells not reporting data Predictive failure analysis Real-time Customer Experience Real-time dropped call detection and resolution for VIP & Corporate customers VIP track and trace across the network Real-time experience for roaming VIPs Real-time Fraud & Security Mobile originated spam detection Mobile wallet fraud detection 2014 www.vitria.com 26
What Problems Can I Solve with Streaming Analytics? Real-time Marketing Real-time 1-to-1 marketing based on where the customer has been, where the customer is, and a prediction of future behaviour Travel-related roaming offers Real-time Revenue Optimisation Real-time mobile data pricing Analytics supporting new account / service propositions such as Joint Accounts Dynamic top-ups and dynamic charging based on usage Real-time Analytics for Internet-of-Things/M2M Operational analytics for Smart Metering, Asset Management, ehealth, Connected Car 2014 www.vitria.com 27
Case Study: Telefonica O2 UK Operational Intelligence for Real Time Customer Experience Problem: Maintain Competitiveness by Maximising Customer Experience Competitive pressures make providing the best customer service essential Demands a shift from traditional service assurance to real-time, 1-to-1 customer focus Requires ability to process huge volumes of data from the network in real time Filter, correlate and enrich events of interest, visualise and act on them in real time Solution: Vitria Operational Intelligence Platform Vitria OI provides real-time visibility, insight and action across network events correlated with customer, network and device reference data Largest Carrier in UK Subsidiary of Telefonica 7th Largest Telco WW >320 Million Customers Big Data in Motion: 250,000 events/sec. ~10 billion events/day 2014 www.vitria.com 28
Network Visibility vs. Customer Insight: How to ensure the best service for VIP Customers? Source: O2 2014 www.vitria.com 29
Operational Intelligence at O2 #1: Real-time Network Situational Awareness Real-time monitoring of network performance and faults Worst performing cells (dropped calls) Corporate and in-building cell monitoring Cell cluster monitoring Adjacent cell performance Cells under detailed investigation #2: Real-time VIP Customer Experience Monitoring Real time Customer Experience for VIPs and High Value Accounts VIP dropped call detection Tracked customers detailed customer experience tracking Inbound roaming VIP customer experience Automated escalation #3: Real-time, Predictive 1-1 Marketing Real-time, relevant offers based on where the customer has been, where they are now, and where you predict they are going O2 Travel related products to customers about to roam off the UK network (Eurostar, UK airports) Largest Carrier in UK Subsidiary of Telefonica 7th Largest Telco WW >320 Million Customers Big Data in Motion: 250,000 events/sec. ~10 billion events/day 2014 www.vitria.com 30
Real Time Customer Experience How to ensure the best service for VIP Customers? Cellular Network 250,000 events per second Volume of Data Velocity of Data Continuous Monitoring for anomalous events (Dropped calls) CRM Correlate among disparate sources CRM, network data, device DB, Continuous Real-time Analytics # VIPs affected per cell Automated Actions Immediate action based on analytics < 1 sec 2014 www.vitria.com 31
VIP / HVA Call Failure Monitoring Geo-map of call failures by VIP group Categorisation and cause code of call failure(s) List of call failures for VIP group Call performance for VIP group 2014 www.vitria.com 32
Real-time, Situational 1:1 Marketing 10 million passengers/year travel on Eurostar trains via the Channel Tunnel between UK and Europe. Many are O2 customers. Most turn off data roaming just before leaving the UK. Eurostar Opportunity: Text them a great data roaming offer just before leaving UK. Problem: Javelin Trains share UK routes Challenge: How to detect customers on the train? Local Javelin trains share same routes Highways next to train routes & stations 2014 www.vitria.com 33
Real-time, Situational 1:1 Marketing Cellular Network 250,000 events per second Volume of Data Velocity of Data CRM Correlate among disparate sources CRM, Route Correlate location with train route Geospatial (location) context Track & trace passenger over time Ensure that this is a train passenger Correlate with train schedule Only track Eurostar passengers Automate actions Text the roaming offer In-Time between Ashford & Tunnel 2014 www.vitria.com 34
Real-time, Situational 1:1 Marketing Customer detection alerts SMS notification status Individual customer journey drill-down Route over time visualisation 2014 www.vitria.com 35
Real-time Analytics for IoT: UK Smart Meter Implementation Programme Largest M2M project in the world 23 million customer hubs, 53 million smart meters across the southern half of the UK by 2020 Vitria OI will provide real-time operational management Power outage management, fraud detection, device troubleshooting Coordination of process-based automated responses Operational analytics 2014 www.vitria.com 36
Summary STREAM Continuously ingest massive volumes of events and data DISCOVER Discover exceptions, patterns and trends ANALYZE Correlate, analyze, and predict outcomes ACT Respond proactively. Seize Opportunities. Squash threats. Vitria Operational Intelligence is a streaming analytics platform consistent with TMF Big Data reference architecture Complementary with Big Data at Rest in Lambda Architecture Combines real-time streaming, discovery, analysis, visualisation and action Scalable, high performance at extreme event rates Enables innovation across multiple operator business silos 2014 www.vitria.com 37
Operational Intelligence Powered by Streaming Analytics Try our Interactive Streaming Big Data Demos vitria.com/big-data-demo www.vitria.com 2014 www.vitria.com 38
inform innovate accelerate optimize Q & A #streamingbigdataanalytics Webinar sponsored by: 2013 TM Forum 39 V2013.4