Filling the Void of Big Data Structure

Size: px
Start display at page:

Download "Filling the Void of Big Data Structure"

Transcription

1 Filling the Void of Big Data Structure Rome, October 17 th, 2017 Alessandro Stagni Sr. SW Product Manager

2 Here We Are Again: What Is Big Data? Big Data technologies and related processes are a reality. Often this ecosystem is perceived as a cool way to implement Data Warehouse. It is all about a bigger Data Warehouse? 2

3 The 3 V s Of Big Data High Volume High Velocity High Variety 3

4 The Big Data Long Tail Generate Data Aggregated Data: Traffic, weather, occupancy, speed, road temperature, Not Aggregated Data for Each Vehicle: Identity, speed, dimensions, make, model, occupancy, Financial Information Toll transaction, transaction status changes, payments, End User Interactions Every click on self-service app, interaction with CSR, device used, Infrastructure Operational Events: Alarms, Work Force Events, Signage, Logs 4

5 Big Data Is Not Only Batch. Events Rules. Stream Data Generate Data Big data information units are Events Events are immutable entities. Events must be gathered, and made immediately available to the processing infrastructure Data Integrity must be implicit A paradigm shift is needed from the architectural point of view to stream all the generated events Third party events can be streamed as well Traffic Weather Interoperability 5

6 Forgetting Nothing Store Data Stream Data Storing is the base for every further activity Tolling processing has the capability to generate hundreds of distint events per single Toll Transaction Cost per Gigabyte in the distribution cluster, and scalability are essential characteristics to take into account Generate Data 6

7 Here Comes The Disruption: Near Real-Time Processing Process Information Store Data Stream Data Big Data technologies are not just a reporting environment There are two planes of data processing: The traditional batch ETL processing of historical data The near real-time processing of streams (aka Speed Layer) In other words: processing the present events based on past knowledge Generate Data 7

8 Data Has No Value Without An Interpretation Extract Meaning Process Information Store Data Stream Data Generate Data Traditional role for meaningful extraction: Data mining, analytics, simulations for: Operations Optimization, Marketing Management, etc. Adding near-real time event processing: Trip Building, Dynamic pricing calculation Managed Lanes Traffic Management Incident Management 8

9 Joining The Connected Realm Share/Trade Meaning Extract Meaning Process Information Store Data Stream Data Generate Data In this context Near Real Time processing means: giving humans the possibility to interact or condition the business with their behavior Human are connected Sharing events is invariable in the connected era Provide better services experience Optimize revenues Understanding your own role in the ecosystem 9

10 IoH (Humans?) IoV (Vehicles) IoT IoR (Roads) The Value Network 10

11 IoH (Humans?) Connectivity Road Networks IoT IoV (Vehicles) IoR (Roads) VANET? The Value Network 11

12 IoH (Humans?) Connectivity IoT IoV (Vehicles) Agencies, Departments of Transpiration, Municipalities, IoR (Roads) The Value Network 12

13 IoH (Humans?) Connectivity IoT IoV (Vehicles) Agencies, Departments of Transpiration, Municipalities, IoR (Roads) The Value Network 13

14 IoH (Humans?) Connectivity IoT IoV (Vehicles) Agencies, Departments of Transpiration, Municipalities, IoR (Roads) The Value Network 14

15 IoH (Humans?) Connectivity IoT IoV (Vehicles) Agencies, Departments of Transpiration, Municipalities, IoR (Roads) The Value Network 15

16 Do We Need To Wait For CAV? Traffic Data Managing Agency Fares (Near Real Time) Dynamic Price changes drivers behavior Take Decision 16

17 Connected Vehicle Now. Autonomous? Traffic Data Managing Agency Fares (Near Real Time) Dynamic Price changes drivers behavior Take Decision Map Service Suggest Choice Traffic Data 17

18 Filling The Void Real-time big data is a huge technology shift Event streaming and processing Data Science Machine Learning and Deep Learning Real-time big data requires a completely new vision in specifying roadside and back office systems. 18

19 Filling The Void (continued) Fragmentation and lack of standardization weakens the agency s role in the big data IoE (Internet of Everting) value chain. Walled gardens have failed in the past. A change to the interoperability approach is possible. Interoperability should include all kind of data 19

20 Filling the void (continued) Big data technologies will have a huge role in the transformation from Toll Collection to Transportation as a Service. Be prepared to interact in complex value chains Understand possible disintermediation, and the economics behind the big data, real-time processing, machine learning, and other technological disruptions. 20

21 Thank You