DIGITAL DECOUPLING OPENMUNICH 2017 Tobias L. Maier
HI! Tobias L. Maier Technology Architect Emerging Technology Innovation @tobias_maier Photo Niklas Altekamp www.niklas-altekamp.de
THE DISCONTINUITY MODEL THE GREEN ZONE Enterprise IT is in shape and the ability to support business change is good, but does not guarantee it THE YELLOW ZONE Enterprise IT is at risk, is it moving up or down? The ability to support business can still be deceivingly good THE RED ZONE Enterprise IT is limiting business flexibility and puts current and future business performance at risk TIME ORGANIZATION IT MOVES IN THIS MODEL OVER TIME
THE BASIC MOVEMENT PATTERN Cruising, focus on new things Cost efficiency, slow decline A program to fix things Problems, same mess for less, outsourcing, Re-boot (consolidate, replace, multi-year programs) TIME TYPICALLY 10-20 YEARS REPEAT
THE PROBLEM It takes too long to get out of the hole. The cost to get out is very high and the competition can run away during this time of innovation inactivity (years) Not able to recognize when troubles start, suddenly things take longer, and in some cases might be impossible to do TOO LONG TIME
THE 7 ANTI- PATTERNS ACCELERATING THE MOVEMENT
7 ANTI-PATTERNS 1. Consolidation as the only approach, driven by 2. A product alone solves a business problem 3. A perfect architecture and governance 4. Functional end state is reachable 5. Adapt IT systems to meet all business processes requirements 6. Co-dependent relationship between business and IT 7. All old = bad
7 ANTI-PATTERNS CONSOLIDATION AS THE ONLY APPROACH Driven by Need to standardize processes across the organization Need to create more real-time solutions Need to reduce costs in a spaghetti landscape
7 ANTI-PATTERNS A PRODUCT ALONE SOLVES A BUSINESS PROBLEM Relying on someone else having the hammer
7 ANTI-PATTERNS A PERFECT ARCHITECTURE AND GOVERNANCE Ivory tower architecture Focus on standards that don t drive structure e.g. canonical data models
THE 7 HABITS OF SUCCESSFUL IT Continuous navigation and the avoidance of discontinuity points using the following: 1. Layering 2. Appropriate fragmentation 3. Simplification 4. Continuous investment 5. Rigorous focus on quality 6. Employ outcome based KPIs 7. Shift to value based organizations TIME
REALITY CHECK DO YOU NEED FULL TECHNICAL PROTECTION FROM DISRUPTION? OR CAN YOU TAKE A BIT LESS DEMANDING APPROACH, MIXING EXPONENTIAL IT WITH CLASSICAL MODELS? TIME TIME
AUTOMATION DIGITALLY DECOUPLING THE CURRENT IT 1 2 ONLINE THE REST SERVICE PLATFORM 3 NEW REALTIME INTERACTIONS 4 DATA EVENTS 5 INTELLIGENCE BI /A 6 1. Automate using RPA 2. Build out differentiated services using Microservices, DevOps and the cloud and provide API access to the common core 3. Utilize Cloud to quickly build new core services 4. Build out a data lake with real-time eventing capabilities 5. Build systems of intelligence 6. move beyond the shopping cart, build for example voice and chat based online services + Seek combinatorial effects + Maximize cloud for new B2B PARTIAL CLOUD
EVOLVE TOWARDS EXPONENTIAL IT THROUGH DIGITAL DECOUPLING AND MODERNIZATION DECOUPLE EVOLVE TOWARDS EXPONENTIAL IT MODERNIZE Copyright 2017 Accenture. All rights reserved. 14
SOMETHING ELSE VS. DECOUPLING Vision and aim of a perfect architecture, the big shot Always evolving architecture allowing to pivot anytime Strict 3/4/5 tiers á la Data Integration Processes UI Purposeful fragmentation with high independence of systems Central integration solutions, e.g. huge ESB Workflow Engines Each part follows simple, open integration standards Consolidation: One canonic data model, one data owner, one * Pragmatic data duplication when needed with simple principles
Internal External ARCHITECTURE EXAMPLE User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Legacy Services Transactions
ARCHITECTURE STRATEGIES TO DIGITALLY DECOUPLE #1: Wrap the Monolith(s) #2: Make data accessible #3: Real-time interactivity #4: Be cloud native #5: Be relevant
#1 WRAP THE MONOLITH LOVE IT CHANGE IT KILL IT What? API Wrapper around Monolith, making it a normal component Encapsulate with API s & clear contracts Standard JSON/XML, but not a beauty contest Products (integration, RPA,...) help a lot Why? Regain control & visibility Opens for a range of future tactics: slicing right-sizing replacing migrating
Internal External #1 WRAP THE MONOLITH LOVE IT CHANGE IT KILL IT User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Legacy Services Transactions
Internal External #1 WRAP THE MONOLITH LOVE IT CHANGE IT KILL IT User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Transaction Svc. REST Legacy Services Transactions
#2 MAKE DATA ACCESSIBLE THE BLOOD IN TODAYS BUSINESS What? Build a working real-time ODS (often an operational data lake) Get all data there, use case by use case Get original data formats, transform within lake Why? Quick & cheap (read) access to data, easy to add services Analytics and AI-driven use cases need data. Lots of up-to-date data.
Internal External #2 MAKE DATA ACCESSIBLE THE BLOOD IN TODAYS BUSINESS User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Transaction Svc. REST Legacy Services Transactions
CDC Replication Internal External #2 MAKE DATA ACCESSIBLE THE BLOOD IN TODAYS BUSINESS User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Transaction Svc. Transactions REST Legacy Services Transactions
#3 REAL-TIME INTERACTIVITY THE BACKBONE OF CUSTOMER EXPERIENCE What? Establish a scalable and agile event processing capability Be able to plug into a data feed to generate new representations and ACT! Why? Drive out AI in real-time; next-bestaction, smart pro-active service, Create real-time mashups of data, new representations (e.g. full-time search), drive automation
CDC Replication Internal External #3 REAL-TIME INTERACTIVITY THE BACKBONE OF CUSTOMER EXPERIENCE User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Transaction Svc. Transactions REST Legacy Services Transactions
CDC Replication Internal External #3 REAL-TIME INTERACTIVITY THE BACKBONE OF CUSTOMER EXPERIENCE Open Source at https://github.com/accenture User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Reactive Information Gateway (RIG) Transaction Svc. Transactions REST Legacy Services Transactions Event Hub Fraud Detection (Analytics) Notification Service Other Utility Svcs. Streaming Messages Synchronous REST Calls
#4 BE CLOUD NATIVE ITS ALL ABOUT SPEED TO MARKET What? Embrace Microservices with clear contracts (e.g. OpenAPI) and simple powerful rules Embed in state-of-the art DevOps procedures, heavily leverage containers Why? Time to market. Teams don t step on each other toes and enjoy controlled freedom Risk-control. A microservice is always replaceable, change-able, upgradeable
CDC Replication Internal External #4 BE CLOUD NATIVE ITS ALL ABOUT SPEED TO MARKET User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Reactive Information Gateway (RIG) Transaction Svc. Transactions REST Legacy Services Transactions Event Hub Fraud Detection (Analytics) Notification Service Other Utility Svcs. Streaming Messages Synchronous REST Calls
CDC Replication Internal External #4 BE CLOUD NATIVE ITS ALL ABOUT SPEED TO MARKET User/Mobile User/Web Customer Support 3 rd Party Services Apple/Google Push API Reactive Information Gateway (RIG) Transaction Svc. User Svc. Process Svc. Transactions User / Prefs. Forms REST Legacy Services Transactions Event Hub New Service New Store Fraud Detection (Analytics) Notification Service Other Utility Svcs. Streaming Messages Synchronous REST Calls
#5 BE RELEVANT YOU GOT THE LOOKS What? Focus on tools that get you forward quickly* Embrace the latest UI/UX Embrace JavaScript Follow the cool-kids (Facebook & co); this is a fast moving field, not architecture perfection Why? Market expectation. Users don t read manuals and want their stuff on all gadgets Taking risks is Ok, lifetime is 2-3 years, key principles and security are ensured on API level * Examples: Portals are often a massive slowdown; React has a nicer learning curve than Angular,...