Snow technical specifications
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
README.md

README.md

What type of data does SNOW collect? (Technical details)

Host telemetry gathering is at the core of SNOW’s approach to defending infrastructure: SNOW has two polling streams for each agent, and each poll at random intervals to reduce their predictability from an attacker’s perspective.

The base module (which handles loading and updating the sensor module) polls with a mean average of 10 minutes;
The sensor module, which acquires the telemetry, polls with mean 2 minutes.

The following data is automatically collected from SNOW; however, please note that any of these collections can be turned off without rebuilding the agent by using dynamic configurations:

  • List of running process, along with path of program, list of included DLLs with paths, and SHA-1 hash of each executable module
  • List of registered services (with metadata)
  • List of loaded drivers (with metadata)
  • List of auto-run programs (with metadata)
  • List of installed programs
  • Time when processes start and end
  • Processes running "cloaked" (e.g. obtained through brute-force polling of all possible PIDs on a system that don't appear in a normal process listing)
  • Metadata (path, size, access times) on files considered suspicious (e.g. program that was running cloaked)
  • Domains to which DNS requests were issued
  • Metadata (IP, port) of UDP and TCP network connections
  • Metadata on files added, removed or modified into directories of interest (exactly which depend on OS and specific client sensitivity)
  • For programs chosen at random, memory pages marked as executable are scanned for the presence of recognizable artifacts of executable code - memory pages are not transferred as telemetry
  • History of IP addresses used by the machine
  • Performance metrics of the agent itself as well as lists of agent execution checkpoints