Keynote: Anything is Possible: Apply Graphs to Your Most Complex Data Problems - Carl W. Olofson
0 likes50 views
The document discusses the capabilities and applications of graph databases, emphasizing their ability to capture complex relationships in data that traditional databases struggle with. It outlines common use cases such as fraud detection, recommendation systems, and logistics, and highlights the future potential of graph databases as they evolve with new technologies. Recommendations for organizations include exploring diverse applications and starting small with graph database projects to fully leverage their capabilities.
Keynote: Anything is Possible: Apply Graphs to Your Most Complex Data Problems - Carl W. Olofson
- 1. Anything is Possible: Applying Graphs to Your Most Complex Data Problems Carl Olofson Research Vice President IDC
- 2. 2 © IDC | Agenda 1 Thoughts on Graph Databases 2 Where Graph Databases Fit in the Database Pantheon 3 Every DBMS Supports Graphs (Not Really) 4 Types of Graph Databases 5 Common Graph Database Use Cases 6 Other Potential Uses of Graph Databases 7 The Future for Graphs, and Database in General
- 3. 3 © IDC | Thoughts on Graph Databases Most DBMSs focus on record-keeping and structured regurgitation. Graph databases can 1 2 Capture the realities of the enterprise based on those realities, not on some modeler’s conception of them, for a “real world” view. Enable discovery of improved efficiencies, hidden issues, better insights regarding customers, products, and operations. Resulting in 1 Lower overall operational cost 2 More effective sales efforts over time with less effort 3 New strategies going forward
- 4. 4 © IDC | Thoughts on Graph Databases Graph databases are the Swiss Army knives of databases. Semantics (document structures) Networks Patterns They can be used for 1 2 3 4 5 Data structured may be pre-defined, or open. Relationship combinations may be anticipated or discovered. A graph database is the perfect platform for data structures generated by the machine learning process. Unanticipated data combinations Complex data combinations
- 5. 5 © IDC | Where Graph Databases Fit in the Database Pantheon Key-value stores are good at rapid storage and retrieval of simple sets of data; well suited for caching and high speed get and put. Document databases are good for isolated applications that don’t share or blend their data, and where the data structure is intimately tied to the application code. Relational databases are well suited to handling data that is easily rendered in two-dimensional structures (tables), and for comprehensive query handling (as in a data warehouse). Graph databases can handle any data structure imaginable and is constrained only by the limitations of the infrastructure.
- 6. 6 © IDC |
- 7. 7 © IDC | Every DBMS Supports Graphs (Not Really) Graphs enable navigation of arbitrarily complex relationship structures. 3 Being able to do a thing doesn’t mean you should do that thing. 2 Any DBMS can store any structure. Therefore, any DBMS can store a graph. 1 No one wants to write graph maintenance and traversal code; it’s just too hard. 6 In a document database or key-value store, data structures must be supported in code. So, if you want to use them to manage a graph, your code is managing the graph. 5 Unlike in a relational database, such navigation is not limited based on a fixed schema. 4 A graph DBMS is specialized for building, traversing, maintaining, and reporting on graphs. 9 A few RDBMS vendors do have specialized functionality that supports graph, but do you want all that other relational overhead just to manage a graph? 8 If your favorite key value store vendor, or document database vendor, or relational vendor says, “we support graphs too,” just smile. 7
- 8. 8 © IDC | Types of Graph Databases Types of Graphs • Semantic Graphs • Knowledge Graphs • Property Graphs 1 Required Functionality • Support for Graph Analysis • Support for Graph Traversal • Support for Graph-Based Transactions 2 Which type do you need? 3
- 9. 9 © IDC | Common Graph Database Use Cases Fraud Detection and Analytics Network and Database Infrastructure Monitoring for IT Operations Recommendation Engine & Product Recommendation System Master Data Management Social Media and Social Network Graphs Identity & Access Management Data Privacy, Risk, and Compliance Artificial Intelligence and Analytics Key Verticals that Use Graph Database Retail Telecommunications Government Life Sciences Financial Services Supply Chain Management Legal Record-Keeping (Case Law)
- 10. 10 © IDC | Examples of Value from Graphs 1 2 3 4 Logistics – Dynamic Routing for Faster and Cheaper Pickup and Delivery Supply Chain Management – Overcome Scheduling Issues Life Sciences – Rapid Drug Discovery Marketing – Identify Shifting Demand Patterns and Sell More Product.
- 11. 11 © IDC | Other Potential Uses of Graph Databases 1 2 3 4 Metadata Management, Including Schema Capture or Discovery Geospatial Data Capture and Analysis Information and Knowledge Management Training Data Management for Generative AI
- 12. 12 © IDC | The Future for Graphs, and Database in General Graph databases will benefit tremendously from a commonly accepted GQL (graph query language). Leading graph DBMS providers are working on a common, SQL-like language. Since the lead inhibitor for graph implementation is hardware, progress in increasing processor speeds while reducing cost is key to broader and deeper graph deployment. There is increasing interest in multi-model DBMS. Most start from a base and expand. Oracle, for instance, expands from a relational base. MongoDB expands from a document base. It could be argued that the best base to start from is a graph base, since it can be adapted to the other formats, but it is difficult to adapt the other formats to it.
- 13. 13 © IDC | Conclusions and Recommendations Graph Databases and Their Future • The IT world has barely scratched the surface of important use cases for graph databases. • Graphs can form a foundation for a range of mixed operations that may include some blend of tables and documents in addition to graphs. • AI/ML will flourish with graph databases as a data platform. Conclusions Recommendations • If you are exploring graph databases, talk to existing users and discover the broad range of possible applications. They go well beyond fraud detection and pattern discovery. • If you are a current user of graph databases, consult your team, and look at all the other problem spaces in your business where limitations on data organization is holding you back. • Don’t be afraid to expand and explore, You may be able to do more than you think with graphs. • Consider graphs for cases where the organizational requirements of data what applications can do. • Start with a small project, and build from there • Learn from the experience of others… there are numerous useful cases online. Getting Started
- 14. 14 © IDC | If you can think it, you can graph it.
- 15. IDC.com linkedin.com/company/idc twitter.com/idc blogs.idc.com © IDC Carl Olofson Email: colofson@idc.com Twitter: @databaseguru LinkedIn: linkedin.com/in/carlolofson