Diting Labs
Talk to us

Explore the system beneath

Diting Labs is building the world's best observability platform.

See the platformTalk to the team

The system is already talking

Diting reads signals from the kernel, the wire and the runtime. No code changes. Metrics, logs, traces, database statements and session replay are time-aligned.

One host, one agent. Install it and look.

OverviewDiting
Alerts, resource views, profile and kernel hotspots, dynamic tracing.
Alerts, resource views, profile and kernel hotspots, dynamic tracing.

A slow query, down to the click

Start from a slow statement in the database and end at the page the user actually saw. No frontend instrumentation, no re-enactment.

  1. Database

    Where the statement hurts

    MySQL, PostgreSQL and Oracle statements and plans are parsed at the protocol layer. No driver involved.

    01
  2. Trace

    Which request owns it

    The statement hangs off a span, with upstream and downstream services, duration and errors attached.

    02
  3. Session

    What the user was doing

    The request is tied to a browser session: clicks, typing, navigation.

    03
  4. Replay

    Watch the recording

    See the page, where it stalled and what it threw, in order.

    04
The scene behind a slow queryDiting
01Slow querySELECT … WHERE order_id = ?
02Owning request
03User session
SQL, request, session and replay, time-aligned.
gatewayworkermysqlredischeckoutedge

What Diting does not do

01

No code changes

No SDK, no agent jar, no manual spans. Vendor software and systems nobody dares to touch are fair game.

02

No in-app probe

Nothing is injected into the application process. No launch flags, no restart.

03

No public network

Install, upgrade, licensing and AI models arrive in the same bundle. The machine can stay offline forever.

04

No platform swap

Kernels from 3.10, domestic distributions, Kubernetes and virtual machines side by side.

Fully offline deployment

Packages, images, dependencies, licensing and AI models ship in one offline bundle, upgrades included.

  • No egress
  • Offline license checks
  • Offline upgrade bundles
  • Local model loading
Offline deliveryDiting

Intranet

  • Packages and images
  • License file
  • Upgrade bundle
  • AI models
no egress

Public network

Not involved
Self-contained: no public network, no impact on install, upgrade or daily use.

AI can act

It can open pages Diting has adapted, filter, drill down and run actions you allow. Permissions and audit follow it.

  • Per-page contracts: each product page declares what AI can read and execute.
  • Permissions first: what it can see and change is set by role.
  • Every action is recorded and searchable in audit.

Commit, build, release, roll back

Jenkins, registries, environments and release records are connected. AI writes the app, then builds and ships it, with observability on from day one.

One pipelineDiting
01Commit
02Build
03Release
04Observe
05Roll back
Commit, build, release, observe and roll back, time-aligned.

Standalone, performance, HA

Standalone, performance and HA use the same packages and upgrade path.

StandaloneThe whole platform on one machine. Evaluation, branch offices, edge sites.
Performance clusterCapture, storage and query scale separately. Larger host and container estates.
HA clusterReplicas, failover, rolling upgrades. Systems that cannot stop.

Start with the first host

Start with host installation and kernel requirements.

Open the docsTalk about your estate