Week 3: IT Resilience Graph – Singapore
GraphAcademy Cup Team Profile Link: Singapore | GraphAcademy Cup
GraphAcademy Public Profile Username: leongkwokhing
Country: Singapore
GraphAcademy Course Completed Course Name: Graph Data Modeling Fundamentals
Project Name: IT Resilience Graph
Description: This project leverages Neo4j Aura to build a dynamic IT infrastructure dependency model that unifies hardware, network, and application layers into a single knowledge graph. By mapping connections between critical internal assets and vendor-provided resources (such as Stripe and Singtel), the graph replaces traditional, siloed asset tracking with interconnected intelligence.
Designed for high-stakes operational resilience, the solution enables IT teams to perform instant blast-radius assessments and proactive risk management. For instance, teams can execute multi-tier variable-length path queries to instantly trace how a hardware failure cascades up to mission-critical applications, or identify how a single external vendor outage (like a Singtel telecom disruption) propagates across organisational boundaries to compromise 8 distinct downstream IT assets.
What Did You Learn?: The course provided a reinforce my understanding of graph theory and graph database architecture. Moving beyond the limitations of rows and columns, I learned how nodes, relationships and properties form an expressive, interconnected web capable of mirroring real-world complexities. The course's deep dive into common graph use cases equipped me with the logical framework needed to model intricate data ecosystems. Furthermore, mastering Cypher pattern matching unlocked the ability to turn raw data points into actionable insights.
I directly applied these concepts to build the IT Resilience Graph, transforming traditional, siloed IT asset tracking into an interconnected intelligence model. By organising hardware, network, and application components as distinct nodes and defining their relationships, I successfully mapped a dynamic topology across internal and external vendor boundaries. Applying the graph modelling and pattern-matching techniques learned in the course allowed me to implement the graph schema variable-length path queries in Neo4j Aura. This capability enables IT teams to perform precise blast-radius assessments during live incidents, tracing how a single localised failure—such as a hardware crash or a Singtel telecom outage – cascades across architectural layers to compromise critical upstream applications.
Screenshot
Subgraph in Neo4j depicting the dependency between IT assets:
[Impact Analysis] If the SAN Storage Array experiences a hardware crash, using
variable-length paths (*) to traverse multiple tiers of infrastructure layers instantly
reveal that both the Customer Database and Oracle Database (and sequentially, the
E-Commerce site and ERP) will be affected.
[Vendor Risk Management] By tracing paths across external boundaries, the graph
reveals that a single Singtel telecom outage triggers a domino effect impacting 8
distinct IT assets. This showcases Neo4j's unique ability to map cross-organisational
dependencies for proactive vendor risk management.
Additional Notes: While path traversal shows what is currently affected, we can explore using Neo4j’s Graph Data Science (GDS) library to calculate hidden infrastructure risks before an outage happens.
For example, running the Betweenness Centrality or PageRank algorithm on the (:Component) nodes. This could pinpoint the most critical infrastructure bottlenecks (e.g., a specific router or database) that have the highest number of shortest paths passing through them. If these nodes go down, the largest chunk of your infrastructure goes down.