Abstract This practitioners' report describes Credit Suisse's new - PDF document

Public Design and Implementation of an agile Price Management Platform for Banking Date: 2011-05-19 Produced by: Thipor Kong Account & Product Management / Domain Architecture, Extended Pricing Platform Ulrich Hildebrand Core Banking Architecture, IT Private Banking Architecture FINAL Abstract This practitioners' report describes Credit Suisse's new client-centric price management platform, with the aim of replacing over 20 heterogeneous older applications with an integrated solution. We plan to improve business agility and to enable new ways of creating, controlling, and governing prices across all the bank's businesses. We have decided to employ the federated architecture pattern as an enabler for a stepwise inside-out renewal. This approach leads to a highly distributed platform, with customizable components for specific business needs. As a result, the platform is designed and managed as a software product line. A software product line leverages reuse potential and helps to enforce required commonalities—but a product line also requires a dedicated design and management approach. We'll present our approach, which is based on model-driven engineering and domain-driven design, and will show how a clean separation and customizability of both domain-dependent and domain-independent aspects are supported. $Id:: XPP_SEI_SATURN_20110519.ppt 6758 2011-05-19 21:17:41Z $ $URL:: https://svn01.csintra.net/svnrep/xpp/trunk/doc/pm/XPP_SEI_SATURN_20110519.ppt $ Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 2

Content � The Federated Architecture of the Extended Pricing Platform � Federated Architecture in the wider Context of Credit Suisse Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 3 Enable innovative price models by building an integrated IT platform for client-centric price management Core Banking Processes Brokerage Payments Custody ... Client-centric Banking Processes CRM Pricing Regulatory Support ... But: The matrix of core banking processes and client-centric banking processes must be carefully organized! Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 4

The disconnected and highly heterogenous structure of legacy pricing functionality has to be overcome 20 Standard 40 services; Product Areas Feeders Tariffs 200 variants Feeders Feeders 500 Mgmt & revenue-related Financial types of posting records; Accounting call 90 consignments Special 50 product groups; Terms & XPP Tariffs 130 products Integrated Pricing Platform But: A monolithic, centralized architecture is not an option! Efficient Business/IT rollouts have to be enabled; product area * all numbers approximate specialities and operational independence have to facilitated. Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 5 While more complex, a Federated Architecture is required to balance client-centric, integrated price management vs. feeder variety/independence and rollout complexity Brokerage Tariff Price Mgmt Calc Straight-through processing Payments Service Tariff Price Relationship Manager, Mgmt Mgmt Calc Service Manager, Product Manager Custody Tariff Price Mgmt Calc But: The Federated Architecture implies a high degree of componentization, resulting in the risk of high SW cost and conceptual/operational fractionation. Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 6

The Software Product Line approach provides the tools to enforce required commonalities and to leverage reuse potential � Core Asset Development – Developing components and accompanying artifacts � Product Development – Developing customized products by composing customized components � Organizational and Technical Management – Keeping everything in check But: Effective reuse can only be achieved by a careful and differentiated analysis of all conceptual and technical aspects of the architecture. Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 7 Example: The value of a proper and properly executed global IT reuse strategy Strategy: "Evolutionary" Strategy: "Disruptive" 3*x mCHF x mCHF 1+3+1 countries + CH 6 countries + CH one common four separate Reorganization / Strategy change code base and irreconcilable for all code bases locations & CH 2006-2008 2008-today Source: IT Plan; classification of projects not validated w/ providers Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 8

The XPP reuse strategy is based on differentiated approaches to price model management and price calculation data-driven logic-driven Applications Service & Product & STT Price Contract Management Calculation Management getTariff Business Objects ServiceArea, ProductCombinations, ServiceGroup, Service, Products, ServiceAssignment Tariffs, STTs (Contract ) Universal Feeder-specific Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 9 The common concepts for structuring integrated price models across all product areas XPP BOM Investment Services Service Group (simplified) Discretionary Service Category Mandates SMWB Benchmark oriented "John Doe" holds an "John Doe" Service Area Mandates Exclusive Selection Mandate Service Customer ExclusiveSelection Service Assignment InformationFile Special Tariff Terms of "John Doe" Brokerage Product Area (if applicable) STT TMWB Equities, Warrants, Product Specialisation etc. 0% (no Transaction Fees for 1.6% of Transaction Tariff Exlusive Selection) Value Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 10

Structural Variability: Generic price model structures can be specialized for product areas/feeders If we support variety in the data models – how to efficiently implement the business layer for managig the data? Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 11 Compositional CRUD: A balanced approach for complex data management Data Use Case oriented oriented Compositional CRUD � Arbitrarily complex graph transformations can be represented as composition of basic transformations � Compositional CRUD: The client composes complex operations from primitive operations. The composition can be executed w/ one service call Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 12

Model Driven Engineering facilitates efficient implementation of data and service layers for price model maintenance Conceptual Logical Physical PDM BOM LDM (declarative historization) generated generated (model-to-code) (model-to-model) manual Historized LDM-H (explicit historization) (model-to-code) Access Support generated uses generated JPA (model-to-model) Persistence Unit Resource Access Layer (Java) generated (by JPA provider) LDM-O (optimized) (model-to-text) DB Schema generated Definition (DDL) generated (model-to-model) Supplemental DDL specifications Resource Layer (DB) Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 13 The Economics of MDE XPP TMWB Brokerage FrontNet FNAPPL LDM 27 90 Classes LDM-H 50 149 Generated Artifacts TransferObjects 11 52 kLoC PersistentObjects 2 8 Services 6 21 Generators and Framework M2M 2.5 kLoC TransferObjects 1.5 PersistentObjects 0.1 Services 0.5 Runtime Framework 3* Produced by: Thipor Kong, Ulrich Hildebrand * further refinements steps pending Date: 2011-05-19 Slide 14

Step-wise refinement and variablity of pricing logic XPP Core Domain Logic + Specific Rules Formalism Test Infrastructure Production (Persistence Unit, + Brokerage Specific Logic Infrastructure JavaDB, (Persistence Unit, Test Orders in XML) Oracle, JMS/MQ) + Brokerage Pricable Brokerage PRICE Unit Test getPrice() Service Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 15 Behavioural Variability: Logic of Price calculation can be specialized as well by applying "Stacked Domain Driven Design" Generic Domain Model Brokerage Payments Custody Domain Model Domain Model Domain Model Domain Infrastructure Brokerage Payments Custody Infrastructure Infrastructure Infrastructure Binding Binding Binding Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 16

Distribution of Cost in XPP Architectural Release 1 Specific for Reusable Base first Feeder Component 10% 90% SMWB XPP Platform Components 20% 80% TMWB 60% 40% PRICE & Migration Feeder Integration 70% 30% Integration 80% 20% Price Model Migration 45% 55% (33% of platform) (66% of platform) Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 17 Indicative Reuse Base plus Project Cost & Savings Feeder Total Savings due to base Project Specific cost Project reuse potential Feeder 1 Feeder 2 Feeders 3-5 Feeder 6 Feeder 7 Total Reuse Total Base Savings The more feeders integrated... � the higher the base of reusability � the lower the cost of every subsequent feeder ( � higher cumulated savings) � the higher the relative impact of integration and migration cost ('renewal') Produced by: Thipor Kong, Ulrich Hildebrand Date: 2011-05-19 Slide 18