View source

From BiCEP

== Functional Requirements == * Creation of (base) events ** Procedures/applications ** Declarative based on temporal data model * Event operations (selection/fusion/enrichment) ** Selection ** Projection Composition Enrichment (complex) pattern recognition maybe based on regular expressions) Subscriptions and dissemination Event definition and selection Publisher/data source selection (fixed or attribute based) Consumer selection (fixed or attribute based) Mutual filtering (yes/no) Rich visualization (e.g., dashboards) Need more stuff? . Advanced analytics . Is this scenario distributed in nature? (Yes/no) . This should be integrated with the event operations Have out-of-order events? (Yes/no) . Uses historical analysis? (Yes/no) . Size of moving windows (very big/big/small) . Similarity search (i.e., find patterns that look like this pattern) (precision/recall?) Operational requirements . Throughput (Ultra High throughput/High throughput/Medium throughput) . Response time (>10 sacs/>1 sec/sub-second) . Enrichment performance . Scalability (i.e., how big is the problem in terms of data size, queries/users). Record size . Benefit from computational sharing (i.e., if there are many similar queries that can be processed more efficiently jointly by computing the common part than by treating each query separately) . What is the complexity of the detection/pattern matching logic? (Very high/high/medium) . Provides transactional guarantees? . Privacy . What is the relative cost of false positives and negatives? (High/low) . What value-add does this solution provide? (End-user productivity/new types of applications) . Amount of work to integrate EP with existing systems? (High /medium) . Security . Audit . Tracking . Ad-hoc querying and adaptively Product TOC Requirements . Cost of SW/HW . Education . Development cost/ease (e.g., API available?) . Deployment . Maintenance/Migration . Interoperability . How good is the product scalability in terms of? Scale-up: increase the system and increase the load Speed-up: increase the system and maintain the load Load-up: maintain system but increase the load . Types of communication allowed (push, pull, scheduled)? Domains Business: Algorithmic Trading Business Activity Monitoring Retail Operations Automation Fraud Detection Sensor Driven Health Monitoring Military Ambient Intelligence Traffic Control Nature Monitoring (avalanche, forests, oceans, etc) New Apps Gaming (MMOG) IT Network Management System Diagnostics Big Science Particle Physics Radio Astronomy Mamy more missing…

Return to Use Cases.