<ol> <li>Architecture & Reliability Expectations (Event-Driven Systems) </li></ol> <p>A. Ensure Real-Time Event Pipeline Stability</p> <p>Expected capabilities:</p> <p>Design and enforce idempotent event processing, at-least-once or exactly-once delivery semantics</p> <p>Currently Duplicate detection is enabled at Azure ServiceBus Infrastructure level, and at Application-level messages being validated against duplicate processing.</p> <p>Manage dead-letter queues (DLQ) and retry policies</p> <p>Define event ordering, deduplication, schema evolution</p> <p>Maintain SLOs for latency, throughput, and event loss. Define the same for new designs coming up for upcoming Transaction Groups.</p> <p>Expected actions:</p> <p>Event backlog monitoring</p> <p>Consumer lag thresholds</p> <p>Circuit breaker & backpressure strategies</p> <p>Success metric:</p> <p>Event loss = 0</p> <p>Reprocessing success rate > 99.9%</p> <p>End-to-end event latency within SLA</p> <ol start="2"> <li>Disaster Recovery & Multi-Region Maturity </li></ol> <p>Since the environment is not yet fully Active-Active, the architect is expected to own the phased maturity roadmap.</p> <p>A. Move Toward Active-Active</p> <p>Expected roadmap ownership:</p> <p>Front Door/APIM routing</p> <p>Stateless compute enablement</p> <p>Geo-replicated messaging (Azure ServiceBus)</p> <p>Cosmos DB multi-region writes</p> <p>Azure SQL Hyperscale read replicas / failover groups</p> <p>Events Rejected & Reprocessing Governance</p> <p>B. Azure Data Factory DR</p> <p>Expectation:</p> <p>Design cross-region failover strategy (Example: EA2 ? Central US)</p> <p>Success outcome:</p> <p>DB2 connection resiliency validated through DR drills</p> <p>C. Azure FunctionApp DR (Example: DB2 Integration)</p> <p>Expectation:</p> <p>Architect geo-redundant Function Apps</p> <p>Ensure:</p> <p>Cross-region deployment pipelines</p> <p>Storage replication (GRS / RA-GRS)</p> <p>Mainframe failover connectivity strategy</p> <p>Document file ingestion failover workflow</p> <p>Success outcome:</p> <p>Functions can fail over to Central US with minimal downtime</p> <p>File ingestion resumes without data loss</p> <ol start="3"> <li>Reprocessing - Failed Messages (Maintain current solution) </li></ol> <p>A. Maintain Reprocessing Framework- Expected deliverables:</p> <p>Event replay pipelines</p> <p>Audit trail for rejected events</p> <p>Data consistency & reconciliation logic</p> <p>Replay throttling & idempotency safeguards</p>