Deployment Profiles

CT-Ops ships with three deployment profiles, each targeting a different scale. All profiles use the same codebase — only the infrastructure backing them changes.

Profile Comparison

Profile	Queue	Ingest instances	Database	Suitable for
`single`	In-process (channels + WAL)	1	PostgreSQL (single node)	< 50 hosts, homelab, trial
`standard`	Redpanda (single node)	1	PostgreSQL (single node)	Most organisations
`ha`	Redpanda cluster	Multiple (behind HAProxy)	PostgreSQL primary + replica	High availability required

Single Profile

The simplest deployment. Everything runs on one machine in Docker Compose. The queue is implemented with Go channels backed by a write-ahead log — no Redpanda, no separate consumer processes.

docker-compose.single.yml
services:
  db:       # PostgreSQL + TimescaleDB
  web:      # Next.js, includes embedded consumers
  ingest:   # gRPC ingest service

Use when:

Evaluating CT-Ops
Small infrastructure (< 50 agents)
A single point of failure is acceptable

Standard Profile

Adds Redpanda as a single-node message queue. The metrics, alerts, and events consumers run as separate processes. More resilient — the queue buffers data if consumers are temporarily offline.

docker-compose.standard.yml
├── db          (PostgreSQL + TimescaleDB)
├── redpanda    (single node)
├── web         (Next.js)
├── ingest      (gRPC)
├── consumer-metrics
├── consumer-alerts
└── consumer-events

Use when:

Running 50–500 agents
You want Redpanda's durability guarantees
Single machine is still acceptable

HA Profile

Full high-availability configuration. Multiple ingest and web instances behind HAProxy. PostgreSQL primary/replica pair. Redpanda cluster (3 nodes).

docker-compose.ha.yml
├── db-primary        (PostgreSQL + TimescaleDB)
├── db-replica
├── redpanda-1
├── redpanda-2
├── redpanda-3
├── web-1             (Next.js)
├── web-2
├── ingest-1          (gRPC)
├── ingest-2
├── haproxy
├── consumer-metrics
├── consumer-alerts
└── consumer-events

Use when:

Running 500+ agents
Downtime is not acceptable
You need horizontal scaling

Queue Topic Reference

Regardless of which queue implementation is in use, the topics are:

Topic	Written by	Consumed by
`metrics.raw`	Ingest	Metrics consumer
`events.raw`	Ingest	Events consumer
`alerts.pending`	Alerts consumer	—
`agent.status`	Ingest	Events consumer

Switching Profiles

Profiles are not currently hot-swappable. To move from single to standard:

Take a PostgreSQL backup: pg_dump ct-ops > backup.sql
docker compose -f docker-compose.single.yml down
docker compose -f docker-compose.standard.yml up -d
Restore the backup if needed

Air-Gap Deployment

All three profiles support fully offline deployment. See Air-Gap Deployment for the bundle and transfer instructions.