Telstra Presentation GraphSummit Melbourne: Optimising Business Outcomes with Data Mesh - The Strategic Role of Knowledge Graphs

Telstra Presentation GraphSummit Melbourne: Optimising Business Outcomes with Data Mesh - The Strategic Role of Knowledge Graphs

• 1.
  Optimising Business Outcomeswith Data Mesh The Strategic Role of Knowledge Graphs Senthoor Punniamoorthy Ent Arch Lego Builder-Emerging TechWhiz
• 2.
  General Agenda Page 2 CopyrightTelstra© The Traditional Data Challenge The Path Forward – Data Mesh The Interestedness Challenge Why Knowledge Graphs for Data Mesh Data Mesh Knowledge Plane Knowledge Graph Ontology Standardization
• 3.
  The Traditional DataChallenge The Problem "Throwing Data Over the Fence” Separation between Ops & analytical data ❖ Operational teams manage day-to-day data ❖ Analytical data relegated to centralized teams ❖ Creates a disconnected "throw it over the fence" mentality The Pain Friction… Friction… Friction ❖ Compromised data quality ❖ Delayed data availability ❖ Underutilized data assets ❖ Limited value extraction
• 4.
  The Software Industry'sPainful Past Before DevOps ❖ Development teams built code, operations teams ran it ❖ Teams working in silos ❖ Slow deployment cycles ❖ Production incidents nobody could solve quickly The Answer: DevOps Revolution ❖ Made development teams accountable for both building AND running applications ❖ Pattern repeated successfully with: • Security (DevSecOps) • Testing • Infrastructure management No amount of tooling, processes, or governance could solve a fundamentally flawed operational model
• 5.
  The Path Forward– Data Mesh Addressing the Root Cause ❖ Moves accountability for analytical data to where domain knowledge exists ❖ Eliminates the painful "throwing over the fence" culture Balance of Federation and Excellence ❖ Domains own their data products and outcomes (Domain Driven Design) ❖ Central team transforms from bottleneck to enabler: • Enterprise-wide frameworks / Reusable self-service platforms • Governance guardrails / Security and privacy compliance Data-Driven Excellence ❖ Positions organizations to thrive in the age of exponential data growth ❖ Transforms raw information into actionable business value at speed ❖ Scales value extraction across every business dimension This isn't just a path forward—it's the only viable path for organizations seeking to thrive in an era of data abundance.
• 6.
  The Interconnectedness Paradox Integratedview breaks down silos and enables broader analysis of relationships, dependencies, and organisation-wide patterns that individual operational datasets cannot provide. Analytical data draws its strength from its integrated nature, joined across many different Domains Domain teams own their data products How to maintain cohesion in a world where accountability is distributed?
• 7.
  The Interconnectedness Challenge Pre-matureOptimization ❖ Fixed schemas defined upfront ❖ Rigid data models ❖ Heavy upfront integration planning The Last Responsible Moment Principle ❖ Delay decisions until you have maximum information ❖ Keep options open as long as possible ❖ Make decisions when they're necessary, not before Distributed while interconnected at Runtime? How?
• 8.
  Why Knowledge Graphsfor Data Mesh Distributed Type System ❖ Each data product defines its semantic types ❖ Types can reference other data products ❖ Knowledge graph maintains relationships Global Identification ❖ URI-based addressing ❖ Globally unique identifiers ❖ Resolution services Semantic Linking ❖ Context-aware connections ❖ Temporal awareness ❖ Relationships carry rich metadata enabling runtime composition
• 9.
  General PowerPoint Template Page 9Copyright Telstra© Knowledge Graph Metadata Framework for Data Mesh Knowledge Layer Runtime Composition Use Composition Define how data products can be combined and filtered Guide valid join strategies, Manage cardinality relationships, Enable dynamic data product composition, Apply filtering constraints and rules, Handle set-based operations Semantic Define business meaning and relationships Guide valid compositions across domains, Ensure semantic consistency, Apply business rules during composition Structural Define data shapes and relationships Map fields across data products, Handle composite keys, Manage transformations Quality & Integrity Define data quality and integrity aspects Apply variance tolerances, Check completeness, Validate referential integrity, Enforce uniqueness constraints, Maintain data consistency, Handle null validation Temporal Define time-based aspects Align different time granularities, Handle historical data, Manage effective dates Performance & Distribution Define optimization and distribution patterns Guide query optimization, Manage data distribution, Handle skewness, Apply access patterns, Enable resource optimization, Scale operations efficiently
• 10.
  General PowerPoint Template Page 10Copyright Telstra© Detailed Specification for Runtime Composition Intelligence Metadata Category Elements Runtime Implementation Join Specifications Cardinality Information Guide join strategy selection Join Key Compatibility Validate join conditions Join Type Preferences Apply optimal join methods Cross-Domain Join Rules Enforce domain join policies Filtering Intelligence Valid Value Sets Validate filtering conditions Set Membership Rules Guide set-based operations Scale & Unit Alignment Unit Conversion Rules Apply automatic conversions Scaling Factors Adjust scales during joins Precision Requirements Handle rounding rules Temporal Composition Time Window Rules Align temporal data Granularity Mapping Normalize time units Quality Controls Quality Thresholds Validate quality criteria Consistency Rules Apply consistency checks Nullability Rules Guide null-aware joins Default Values Apply null substitutions Missing Value Strategies Manage missing data Security, Governance, Access Patterns Access Controls Enforce access policies Privacy Requirements Apply privacy controls Usage Context Adapt to usage scenarios
• 11.
  Data Mesh KnowledgeGraph ❖ All functions Data Mesh Powered by Knowledge Graph ❖ Knowledge Plane Multi – Layer Abstraction ❖ Schema Evolution and Extensibility ❖ Open and Standard Based ❖ Machine and Human Interoperability Data Product Life-Cycle Management Data Contracts with Automated Validations Computational Governance & Security Data Mesh Health Observability / Usage Analytics Discovery and Subscription Management Play Ground Metadata Management Self Service & Developer Experience Composition & Dynamic Query Generation
• 12.
  Challenges in DataMesh Implementation Operational Constraints ❖ Limited ability to adopt best-of-breed solutions ❖ Dependency on vendor-specific roadmaps ❖ Higher integration costs ❖ Extended implementation timelines Implementation Challenges ❖ Complex point-to-point integrations ❖ Inconsistent metadata models ❖ Redundant functionality ❖ Performance bottlenecks ❖ Scaling limitations Fragmented Landscape ❖ Vendor-specific implementations ❖ Proprietary knowledge representations ❖ Limited interoperability standards ❖ Siloed tool ecosystems Long-term Implications ❖ Vendor lock-in ❖ Reduced innovation capability ❖ Higher total cost of ownership ❖ Limited architectural flexibility ❖ Impaired digital transformation
• 13.
  Standardizing Data MeshKnowledge Graph Ontology Call for Industry Standardization Key Benefits ❖ Vendor-neutral implementations ❖ Plug-and-play tool integration ❖ Reduced integration costs ❖ Innovation acceleration Implementation Framework ❖ Reference architecture ❖ Standard APIs and protocols ❖ Validation tools ❖ Compliance certification For Vendors ❖ Larger Market opportunity ❖ Faster innovation ❖ Clear integration paths ❖ Focused differentiation For Organizations ❖ Freedom of choice ❖ Best-of-breed solutions ❖ Future-proof architecture ❖ Reduced vendor dependency