jvmguard

Security and data flow

What jvmguard does inside your JVM

This page states what the jvmguard agent collects in your JVM, where it sends that data, and what you can check in the source. For a tool that runs in production, that is the first question to answer.

§ 01Architecture
jvmguard architecture: agents in monitored VMs connect to a server that runs the collector, an embedded H2 database, and the web UI.
Agents push to your server over an authenticated connection. Users reach the web UI over HTTPS. No component talks to a vendor.
§ 02Collected by default

jvmguard records only lightweight data, at low steady-state overhead:

§ 03Captured on demand

Deep artifacts are not collected by default. A capture runs only when a trigger fires, or when you start one manually on a single JVM.

§ 04Operational posture

No phone home

Agents connect to your jvmguard server. The server makes no outbound calls. All data stays inside your network.

No JMX port

The agent reads MBeans in process. You do not have to open a JMX connector server on the monitored JVM to inspect or operate on MBeans.

Single H2 file

Recorded data and configuration both live in a single embedded H2 file. There is no external database to operate.

TLS for agents

Agent-to-server traffic can be encrypted and authenticated via TLS.

Three access levels

Admin, Profiler, and Viewer roles partition a single server for separate teams.

Local, LDAP, or SSO

Authentication uses locally configured accounts, LDAP, or SSO, all configurable in the UI.

§ 05License and deployment

Apache License 2.0. Sources are available. Read it, build it, and audit it with your own tooling before it touches production.

Self-hosted. Runs behind your firewall. Suitable for air-gapped networks. No outbound cloud dependency.

Single server. The collector, the embedded database, and the web UI run in one process. No separate services to secure.