VeloDB Demo

End-to-end demo of Apache Doris 4.1.0 covering three workloads — real-time e-commerce analytics, OpenTelemetry observability, and multimodal RAG — packaged as a single docker compose up.

┌─────────────────────────────────────────────────────────────────────────────┐
│  E-commerce dashboard  ──►  backend (Go)  ──►  Apache Doris 4.1 (FE + BE)   │
│   (5173)                    (8081)              (9030 mysql, 8030 http)     │
│                                │                                            │
│                                ├──►  PostgreSQL  (5432)                     │
│                                └──►  Redpanda    (9092)                     │
│                                                                             │
│  Grafana (33000) ───────►  OTEL collector  ◄─── telemetry-generator         │
│   - VeloDB Demo dashboards     │                                            │
│   - Doris App plugin           ▼                                            │
│                          Doris (demo_observability)                         │
│                                                                             │
│  RAG UI                ──►  rag-api (8000)  ──►  Doris (rag_db)             │
│   - chunk explorer          - hybrid search          - 411 chunks           │
│   - knowledge graph         - chat (OpenRouter)      - 300 entities         │
│   - 1024-d BGE-M3                                    - 19 demo images       │
└─────────────────────────────────────────────────────────────────────────────┘

Quick start

git clone <this-repo>
cd velodb-demo
cp .env.example .env                # set OTEL_TIMEZONE if not in PT
docker compose --profile all up -d  # ~90 s for OTEL schema + RAG ingest

Then open:

URL	What
http://localhost:5173	E-commerce dashboard
http://localhost:5173/rag	RAG / GenAI demo
http://localhost:33000	Grafana — VeloDB Demo folder
http://localhost:33000/a/doris-app	Doris-App (Kibana-style log explorer)
http://localhost:8000/docs	RAG API Swagger
mysql -h 127.0.0.1 -P 9030 -u root -pVeloDB@demo123	Direct Doris SQL

Stop with docker compose --profile all down (keeps data) or down -v (wipes volumes).

Profiles

Profile	Adds	Use when
(default)	core e-commerce + Doris + Postgres + Redpanda	quickest demo
--profile observability	Grafana + OTEL collector + telemetry generator + partition init	OpenTelemetry / Grafana story
--profile rag	rag-init + rag-api with bundled 411-chunk corpus	RAG / multimodal story
--profile all	everything	full pitch

What's inside

velodb-demo/
├── docker-compose.yml             ← single source of truth
├── .env.example                   ← tunables (ports, creds, image tags, timezone)
├── docs/TIMEZONE.md               ← Doris-App timezone setup (read this!)
├── doris/                         ← FE/BE config + schema bootstrap SQL
├── postgres/                      ← OLTP schema + TTL job
├── grafana/                       ← provisioning + 4 dashboards (logs/traces/metrics)
├── otel/                          ← collector config + telemetry-gen definitions
├── rag/
│   ├── init/                      ← rag_unified schema + 3 materialized views
│   └── data/
│       ├── chunks.json            ← 411 chunks across 15 docs (BGE-M3, 1024-d)
│       └── images/                ← 19 demo images
├── scripts/                       ← bootstrap + buildx helpers
└── images/                        ← Dockerfile build contexts
    ├── backend/                   ← Go source
    ├── frontend/                  ← React/Vite source
    ├── grafana-image/             ← Grafana base + doris-app plugin (~50 MB)
    ├── telemetry-generator/       ← Eclipse Temurin + Cisco fat JAR
    └── rag-api/                   ← Dockerfile only (build context: ../AI repo)

The compose file pulls pre-built images from Docker Hub by default (namespace velodb/), so a clone + docker compose up works without building anything locally.

Building / publishing the images

The five images are:

Target	Image	Approx size (arm64)
backend	velodb/velodb-demo-backend:4.1.0	~25 MB
frontend	velodb/velodb-demo-frontend:4.1.0	~50 MB
grafana	velodb/velodb-demo-grafana:4.1.0	~660 MB (Grafana + plugin)
telemetry-generator	velodb/velodb-demo-telemetry-generator:4.1.0	~313 MB (JRE)
rag-api	velodb/velodb-demo-rag-api:4.1.0	~295 MB (Python)

The first four build from sources in this repo:

export DOCKERHUB_TOKEN=dckr_pat_...     # or write to ~/.docker/velodb-pat
./scripts/docker-login.sh
./scripts/build-and-push.sh             # all 5
./scripts/build-and-push.sh --no-push   # validate locally without pushing
./scripts/build-and-push.sh frontend    # one image

The fifth (rag-api) has Dockerfile + requirements.txt only — the Python source (hipporag_doris/) lives in the velodb AI repo:

RAG_REPO=/path/to/AI ./scripts/build-and-push.sh rag-api

Multi-arch (linux/amd64 + linux/arm64) via docker buildx.

Timezone gotcha

The Grafana Doris App plugin formats time-range queries in your browser's timezone. Set OTEL_TIMEZONE in .env to match your browser zone (or set the laptop OS to UTC). The included otel-partition-init service handles the historical-partition backfill automatically. Full explanation and per-region examples in docs/TIMEZONE.md.

RAG features that need an API key

The RAG hybrid search and chat endpoints embed user queries via OpenRouter. Add a key to .env (free tier from https://openrouter.ai) and restart:

echo 'OPENROUTER_API_KEY=sk-or-v1-...' >> .env
docker compose restart rag-api

Without a key, the corpus browser (/api/v1/chunks), entity graph (/api/v1/knowledge-graph), and image serving (/api/v1/images/{path}) still work — only semantic/keyword search and chat require it.

License

Apache-2.0 — see LICENSE.

The bundled Grafana Doris App plugin is © SelectDB and ships under its own terms.