See
Metrics, logs, traces, profiles, replay, databases, Kubernetes and kernel events from one capture layer.
See it, then run it
Reads signals from the kernel and the wire. No code changes, no in-app probes.
Start with one host. No re-architecture required.
- L7 protocol families
- 18
- Kernels supported
- 3.10+
- Fault actions
- 36
- Deployment shapes
- 3
Capture pathDiting
Host
ApplicationNo SDK
RuntimeGo / JVM / OpenSSL
KerneleBPF
NICL2 to L7
Diting Agent
Parsed locally- Protocol ID
- Request correlation
- Compress and ship
Signals, relationships, actions, delivery
Metrics, logs, traces, databases and replay do not need separate integrations; topology and actions use the same dataset.
- Observe: metrics, logs, traces, profiles, replay, databases, Kubernetes, kernel
- Relate: topology, impact paths, service flow, flame graphs
- Act: alerts, batch runs, file distribution, remote terminals, chaos
- Ship: CI/CD, images, environments, releases, rollbacks, AI apps
Capabilities
Run
Topology, impact paths, alerts, batch runs, file distribution, remote terminals and chaos.
Ship
Jenkins, registries, environments, applications, pipelines, releases and rollbacks.
Build
AI writes the app, builds and ships it in one pass. Observability from day one.
Platform specs
| Deployment | Standalone / performance cluster / HA cluster |
|---|---|
| Data path | agent → gateway → Kafka → collector → ClickHouse → core → web |
| Ingest | eBPF, OTLP, OpenAPI |
| Access | Roles down to pages and buttons, full action audit |
Fits
- Physical hosts, virtual machines, containers and Kubernetes together
- Kernels from 3.10 and domestic distributions
- Sites with no public network at all
Doesn't fit
- Estates where no agent may be installed on any host
- Business events that only exist if you change application code
Bring your system in
Tell us what you run: OS, kernel, public network or not, virtual machines or clusters. We will map the path in.