Location Analytics - Real-Time Geofencing using Kafka
summary
Prototype implements real-time geofencing by ingesting GPS streams into Kafka, enriching with KSQL UDFs with GeoTools (Java), and emitting enter/exit events; naive cross-join fails, so a geohash sharding step reduces checks and raises scalability. Bulk-UDF returns lists (low latency, medium scale), while a Tile38 integration via Tile38 + Kafka Connect offloads spatial queries and can publish hooks back to Kafka. Proof-of-concept works; needs performance tuning and a source connector for full round-trip.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:38Guido Schmutz
- 2:58Geofencing - What is it?
- 4:41Geo-Processing
- 6:34Apache Kafka - A Streaming Platform
- 8:02High Level Overview of Use Case
- 9:27KSQL - Overview
- 11:27KSQL - Streams and Tabless
- 13:05How to determine "inside" or "outside" geofence?
- 16:14Using Cross Join
- 16:58INNER Join
- 19:16Geofences aggregated in one group
- 24:30Geo Hash for a better distribution
- 25:52Geo Hash Custom UDF
- 30:154 Geofences aggregated by GeoHash
- 36:24Tile 38 - How does it work?
- 39:112 Using Custom Kafka Connector for Tile38
- 40:39Arcadia Data
- 41:56Outlook
related talks
58mRealtime Analytics Geofencing Kafka
31mFun with Sensors
49mEnriching Kafka Applications With Real-time Contextual Data
46mSmart Transit: Real-Time Transit Information with FLaNK
39mAnalysing Public Kafka Data from NASA Satellites
34mChasing the Stream Processing Utopia
57mTaming IoT Data: Making Sense of Sensors with SQL Streaming
47mBuilding a Real-Time Analytics Database