Also available as:
loading table of contents...


Packet capture (pcap) is a performant C++ probe that captures network packets and streams them into Kafka. A pcap Storm topology then streams them into HCP. The purpose of including pcap source with HCP is to provide a middle tier in which to negotiate retrieving packet capture data that flows into HCP. This packet data is of a form that libpcap-based tools can read.

The network packet capture probe is designed to capture raw network packets and bulk-load them into Kafka. Kafka files are then retrieved by the pcap Storm topology and bulk-loaded into Hadoop Distributed File System (HDFS). Each file is stored in HDFS as a sequence file.

HCP provides three methods to access the pcap data:

  • Rest API

  • pycapa

  • DPDK

There can be multiple probes into the same Kafka topic. The recommended hardware for the probe is an Intel family of network adapters that are supportable by Data Plane Development Kit (DPDK).