CERVED e Neo4j su una nuvola, migrazione ed evoluzione di un grafo mission critical dall’on-prem al cloud

Pasquale Fosso, Head of Data Architecture
Luca Cannistrà, Data Architect Specialist

In questo talk racconteremo come abbiamo affrontato la complessità della migrazione in cloud del nostro “mission critical” graphdb, riuscendo a contenere al minimo i disservizi per i nostri clienti ed i tempi di migrazione.

1. Cerved e neo4j su una nuvola, racconto della migrazione di un grafo mission critical dall' on- prem al cloud

2. ION Group Cerved overview ION provides software for the automation of mission- critical workflow and data analytics solutions to more than 100k users among financial institutions, central banks, governments and corporates . Leading provider of trading automation, analytics and infrastructure to the global financial markets, powering key capital markets workflows Leading provider of proprietary capital markets data and software solutions to investment banks, investment firms, advisors and corporates Partner of choice for the digital corporation, enabling automation of workflows across critical treasury and commodities operations Leading provider of core banking and financial software for financial institutions in Italy and internationally Leading provider of credit information and management in Italy and internationally 350 Financial Markets >1m M&A Transactions Analysed $1.2t Cash Balances >2.3k Banking Branches Served >152m Records of Relationships >5m Financial Instruments $4.5t Equity Issuance Executed $2.1t Foreign Exchange Reserves >2.6m End Customers > 8m Company Records >$23t Annual Transaction Value >2m Firm & Individual Profiles 65% Coverage of Revenue of Top 250 Energy Co’s >100m Transactions per Day >30K Customers Access to financial markets Flow automation Compliance Capital structure optimisation Increasing use of data Market digitisation Increasing complexity Increasing regulation Increasing volumes Open banking & digitisation Increasing complexity Increasing regulation Risk mitigation Prioritization mechanisms Client archetype through ML Increasing needs driven by strong tailwinds

3. Cerved Overview Cerved overview What We Do [ functions ] Who We Serve [s egments ] Why We Matter [ domains ] Expertise [ capabilities ] Delivery [ brands ] Credit Management Financial Institutions SMEs Top & Large Corporates Data management ML & AI technologies Advanced Analytics Credit Risk ESG AML Digital Marketing Advisory Credit management Public Administration #1 data ecosystem on Italian Companies & Real Estate Assets 64M payment experiences collected Reference point for int’l institutions (e.g., OECD, IMF) Rating Agency certified by ESMA Risk Intelligence Marketing Intelligence A leading Credit Management service provider in Italy 95%+ of Italian Banks 30k+ companies (1) 150+ 1) Including Micro companies with purchases on Cervedirect.com

4. Cerved overview Being an ESG role model to support Italian sustainability transition Strong commitment Top quality ratings ESG targets in STI Clear roadmap ESG ratings on Cerved Top management remuneration linked to ESG targets reflecting selected SDGs Foster transparency in the system with Independent ESG ratings and assessment Supply chain ESG platform Help companies to change in a positive and sustainable way Share ESG landscape view and understanding Early leader in defining ESG identity in line with int’l best practice A round ESG offering to support sustainability transition of the System

5. Focus – Ecosistema di dati distintivo ed in costante crescita Cerved overview BILANCI VISURE CAMERALI PROTESTI PROCEDURE DA VISURA ELENCO SOCI CARICHE SOCIETARIE DATI UFFICIALI CAMERALI DATI IMMOBILIARI ADDETTI IMPRESE DATI UFFICIALI NON CAMERALI PREGIUDIZIEVOLI ELENCHI SOCI CONSOB DATI PROPRIETARI PAYLINE AML E LISTE RELATIONSHIPS ATECO RIVISITATI GRUPPI NO REA OPEN DATA GARE PUBBLICHE FINANZIAMENTI PUBBLICI FONDO C. DI GARANZIA ISTAT & BANKIT ELENCO START-UP PA NEWS CORPORATE WEBSITE SOCIAL FEED PAGINE SOCIAL AZIENDALI PAGINE WEB WEB/SOCIAL DATA EXPERIAN Credit Bureau Arricchimento con Dati clienti CLIENTI PARTNERSHIPS MOBILITA’ PERIZIE IMMOBILIARI CYBER RISK +40 ANNI DI SERIE TEMPORALI E + 4.000 BUSINESS RULES PER QUALIFICARE E CORRELARE I SET DI DATI

6. State-of-the-arts technologies Cerved o verview «Raw» Data Insight Only «Tech» Business 100% GDPR compliant Data Architectures Blockchain to improve «notarization» processes Natural Language Processing concerns the querying of databases and returning of generated results in visual format 1.000+ Servers (physical and virtual) 600M events monitored annually 40M€+ worth of investments every year 1.1+ PetaByte of stored data Data Lake Partitions dedicated to our clients Artificial Intelligence: comprehensive application which exploits Machine & Deep Learning algorithms and technologies (i.e. xgboost , neural network, tensorflow ) for Data elaboration (i.e. Proprietary score) and Decision-making processes Graph database : dedicated tool for the analysis and effective representation of relationships between subjects (individuals and companies) Cyber Security & Encryption: Encryption data algorithms based on the most innovative methodologies (Secure Multiparty Computation) and Multi Factor Authentication Proprietary Semantic Text Analytics Engine (Dandelion) API Platform with 100+ proprietary APIs Cognitive Ergonomics / Service Design to be applied on company’s processes and products DATA LAKE: State-of-the-Art in Italy DATA ELABORATION: Cutting-edge Advanced Analytics technologies DATA AVAILABILITY & SECURITY: secure and extensive access to data DATA VISUALIZATION: from data to insight for value generation Digital Invoicing: algorithms for the automatic reading and certification

7. Cerved overview Cerved’s proprietary scores are the market benchmarks ECAI = External Credit Assessment Institutions; AI = Artificial Intelligence Note: 1) Including 2,2 Mln Unregistred economics activities Cerved Group Score (CGS) Benchmark credit risk score available on 3 million companies Credit R ating Certified ECAI & Rating Tool for solicited and unsolicited ratings ESG R ating Proprietary methodology to assign ESG ratings and scores Environmental R isk Score Proprietary score based on hydrogeological data of the territory Payline Score Proprietary payment bureau tracking 61m payment experiences Anti- fraud Score Graph-technology powered score integrated with Credit Bureau Growth Score Proprietary score using inter alia companies digital capabilities Collection Score Algorithms that assess and prioritize collection of credit portfolios Real Estate valuation model Proprietary automated valuation model to assess Real Estate values Open banking (PSD2) score AI-based risk score on SMEs & Individuals via checking account data The biggest data graph in Italy providing deep connections among companies, people and real estate 16.7 million people 149 million relationships 119 million real estate assets 8.4 million active companies Proprietary scores & algorithms

8. Need I Dati attorno ad una azienda ACME spa Esponenti Soci Partecipazioni Soci Comproprietà Relazione Affari Le relazioni tra le imprese e le persone rappresentano un grafo naturale costituito da diverse tipologie di relazioni E’ una tipologia di informazione che mette in difficoltà i database relazionali

9. Use case Titolare Effettivo Livello 1: 10% Livello 3: 24.4% Livello 4: 34.2% ACME spa Soc. A (40%) Soc. B (50%) Willy (10%) Soc. A1 (40%) Soc. A2 (60%) Soc. B1 (40%) Soc. B2 (60%) Willy (40%) Duffy (60%) Willy (40%) Soc. B1.1 (60%) Soc. B2.1 (50%) Speedy (50%) Willy (90%) Bunny (10%) Livello 2: 10% Fino al livello n° 3, nessuno penserebbe che Willy esercita un controllo effettivo di maggioranza sulla ACME. Willy sembra un socio minoritario di ACME

10. Capacity Il nostro Neo4j in numeri Name Role Status Neo4j-01 leader online Neo4j-02 follower online Neo4j-03 follower online

11. Architettura On Prem # Operating system RAM (MB) CPU (MHz) CPU type CPU Totali Socket Core Disk (GB) 3 Red Hat Enterprise Linux Server release 7.3 (Maipo) 257653 2400 Intel® Xeon® CPU E5-2640 v4 @ 2.40GHz 40 2 10 2341,888 leader followers

12. Lift&Reshape Replatform Repurchase Architettura Cloud Valutazioni Le strategie di migrazione valutate :

13. Architettura Cloud Lift&Reshape

14. Automation tools Pipeline

15. Upgrade Step 0

16. Upgrade Step 1

17. Upgrade Step 2

18. Upgrade Step 3

19. Open Point: apoc custom Versione 5 Driver Client compatibile con versione 5 Evoluzioni PRO: Fully Managed (Patch, Upgrade…) Scale On-Demand, Without Service Interruption Fully Managed Backups Cross- Region Database Cloning No Downtime

20. Thank You !