NetFlowMonitor

High-throughput NetFlow v9 collector with a multi-threaded processing pipeline.

Pipeline

1. UDP ingest on configured NetFlow port
2. Bounded processing queue + worker pool (processing.workerThreads)
3. Multi-threaded decode directly into IPFlowMetrics rows
4. Async DB writer queue + batch inserts into IPFlowMetrics every 500ms (default)

Config

The app auto-generates config/netflowmonitor.config.json on first start.

{
  "netflowPort": 2055,
  "database": {
    "enabled": false,
    "jdbcUrl": "jdbc:mysql://127.0.0.1:3306/ignite?useSSL=false&serverTimezone=UTC",
    "username": "root",
    "password": "change-me",
    "poolSize": 10,
    "batchSize": 500,
    "databaseQueueCapacity": 20000,
    "flushIntervalMs": 500
  },
  "processing": {
    "workerThreads": 8,
    "processingQueueCapacity": 10000
  }
}

Output

No per-flow console output. The process is optimized for ingest/convert/insert only.

Run (IDE)

Run gg.zema.elizon.netflowmonitor.SFlowMonitor.

Notes

• Backpressure is explicit: both processing and DB queues are bounded.
• Dropped flow counters are logged on shutdown when queues saturate.
• DB writes use HikariCP and JDBC batching for throughput.
• Rows are flushed to MySQL in batches every 500ms by default (or earlier if the batch size limit is reached).

ORIGINAL com.lumaserv.netflow CODE IS OWNED BY https://github.com/LUMASERV/netflow-java AND IS LICENSED UNDER APACHE 2.0