diff --git a/.dockerignore b/.dockerignore
new file mode 100644
index 0000000..c757b87
--- /dev/null
+++ b/.dockerignore
@@ -0,0 +1,11 @@
+/.git*
+/.editorconfig
+/*.example.*
+/config/
+/log/
+/mongo_data/
+/mysql_data/
+/schedules/
+/Dockerfile
+/docker-compose.yml
+/run.sh
diff --git a/.gitignore b/.gitignore
index e1e7224..ca21124 100644
--- a/.gitignore
+++ b/.gitignore
@@ -12,3 +12,8 @@ celerybeat-schedule
 .AppleDouble
 .DS_Store
 .venv
+
+/log/
+/schedules/
+/mongo_data/
+/mysql_data/
diff --git a/Dockerfile b/Dockerfile
new file mode 100644
index 0000000..6f10330
--- /dev/null
+++ b/Dockerfile
@@ -0,0 +1,39 @@
+FROM python:3.9-slim
+
+ARG DOCKER_USER=dawis
+
+RUN apt update && apt install -y --no-install-recommends \
+    neovim \
+    procps \
+    && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
+# Need to downgrade pip to <24.1 since it got stricter with that version and could not install celery anymore
+# Error: Ignoring version 4.4.7 of celery since it has invalid metadata
+#        Please use pip<24.1 if you need to use this version.
+# The first 24.1 version was released 2024-05-06 and the last pipenv version before that was 2023.12.1, which was still
+# bundled with pip < 24.1
+RUN pip install pipenv==2023.12.1
+
+RUN echo "ls -lah \$@" > /usr/bin/ll && chmod 777 /usr/bin/ll
+
+RUN addgroup --gid 1000 ${DOCKER_USER} && \
+    adduser --uid 1000 --gid 1000 --shell /bin/sh --disabled-password --quiet ${DOCKER_USER}
+
+COPY . /app
+
+WORKDIR /app
+
+RUN pipenv install --system
+
+ENV CELERY_UID=1000
+ENV CELERY_BEAT_SCHEDULEFILE_PATH="/opt/dawis/var/beat-schedules"
+ENV CELERY_LOGFILE_PATH="/opt/dawis/logs"
+ENV CELERY_LOGLEVEL="info"
+ENV CELERY_TIME_LIMIT=600
+ENV CELERY_CONCURRENCY=4
+ENV CELERY_TIMEZONE="Europe/Berlin"
+ENV CELERY_BROKER_URL="redis://redis_host:6379"
+
+CMD ["/app/run.sh"]
diff --git a/README.md b/README.md
index bfcf159..9c4b070 100644
--- a/README.md
+++ b/README.md
@@ -43,6 +43,25 @@ Wir unterscheiden bei der Programmierlogik zwischen zwei Ebenen:
 
 Die Grundlage für die Verarbeitung von Daten sind URLs, auf die zugegriffen wird. DAWIS beherrscht sogenannte URL-Sets, d.h. eine Menge an URLs, die gemeinsam behandelt werden kann. So kannst Du zum Beispiel ein URL-Set für wichtige Kategorie-Seiten in Deinem Shop anlegen, oder ein URL-Set für verschiedene Produktdetailseiten. Auf diesen kann man dann gemeinsame Operationen wie z.B. die Prüfung, ob der Canonical-Tag noch vorhanden ist, durchführen. DAWIS kann nicht selbstständig eine Website crawlen und „alle URLs“ behandeln. Das ist auch nicht Sinn und Zweck eines Monitoring- und Alerting-Systems.
 
+## Docker Setup
+
+Um Dawis mit Docker zu starten, sind folgende Schritte notwendig:
+
+* `docker build -t dawis .`
+* `docker compose up -d`
+
+Unter `http://localhost:4321` können Daten in MongoDB eingesehen werden, unter `http://localhost:8090` Daten in MySQL.
+
+Die Konfiguration befindet sich im Ordner `config`. Logs sind im Ordner `log` zu finden.
+
+Beim Starten wird die Datei `delete_me_for_restart` angelegt. Wenn beispielsweise eine Dateifreigabe auf den `config`-Ordner eingerichtet ist, kann Dawis so auf einfache Art und Weise neugestartet werden, nachdem Konfiguration geändert wurde. Dawis empfängt ein SIGINT-Signal und kann kontrolliert seine Prozesse abschließen und neu starten.
+
+Die Ressourcen-Limits können je nach Bedarf in der `docker-compose.yml` erhöht oder weiter eingeschränkt werden, je nach Performance-Anforderungen und Leistung des Servers.
+
+### Debugging
+
+Zum Debuggen kann folgender Befehl ausgeführt werden: `docker compose up --profile debug dawis_debug`
+
 ## Nächste Schritte
 
 Was kommt als Nächstes? Wir sammeln nach dem Lean-Prinzip das Feedback weiter ein. Parallel erweitern wir den Funktionsumfang der vorhandenen Module. Im Fokus stehen diese Bereiche:
diff --git a/docker-compose.yml b/docker-compose.yml
new file mode 100644
index 0000000..7dc8abb
--- /dev/null
+++ b/docker-compose.yml
@@ -0,0 +1,125 @@
+services:
+
+  dawis:
+    container_name: dawis
+    image: dawis
+    environment:
+      CELERY_LOGLEVEL: info
+      REDIS_HOST: dawis_redis
+      CELERY_BROKER_URL: redis://dawis_redis:6379
+    volumes:
+      - ./config:/app/config
+      - ./log:/opt/dawis/logs
+      - ./schedules:/opt/dawis/var/beat-schedules
+    networks:
+      - dawis
+    depends_on:
+      - dawis_mongo
+      - dawis_mysql
+      - dawis_redis
+    healthcheck:
+      test: "(ls config/delete_me_for_restart && test $(stat -c %Y -- /opt/dawis/logs/worker.log) -gt $(($$EPOCHSECONDS - 300))) || kill 1"
+      interval: 5s
+      timeout: 10s
+      retries: 5
+      start_period: 10s
+    deploy:
+      resources:
+        limits:
+          cpus: "1"
+          memory: 1G
+    restart: unless-stopped
+
+  dawis_debug:
+    container_name: dawis_debug
+    image: dawis
+    environment:
+      CELERY_LOGLEVEL: debug
+      REDIS_HOST: dawis_redis
+      CELERY_BROKER_URL: redis://dawis_redis:6379
+    command: ["/app/run-debug.sh"]
+    volumes:
+      - ./config:/app/config
+      - ./log-debug:/opt/dawis/logs
+    networks:
+      - dawis
+    depends_on:
+      - dawis_mongo
+      - dawis_mysql
+      - dawis_redis
+    profiles:
+      - debug
+
+  dawis_mongo:
+    container_name: dawis_mongo
+    image: mongo:4.2
+    environment:
+      MONGO_INITDB_ROOT_USERNAME: root
+      MONGO_INITDB_ROOT_PASSWORD: dawis
+    volumes:
+      - ./mongo_data:/data/db
+    networks:
+      - dawis
+    deploy:
+      resources:
+        limits:
+          cpus: "1"
+          memory: "1G"
+    command: --wiredTigerCacheSizeGB 1
+    restart: unless-stopped
+
+  dawis_mysql:
+    container_name: dawis_mysql
+    image: mariadb:10.6
+    environment:
+      MARIADB_ROOT_PASSWORD: dawis
+    volumes:
+      - ./mysql_data:/var/lib/mysql
+    networks:
+      - dawis
+    restart: unless-stopped
+
+  dawis_mongo_gui:
+    container_name: dawis_mongo_gui
+    # you need to build this image from https://github.com/arunbandari/mongo-gui yourself
+    # check out 82550d257650c72deac8a19007a9da83ca685423, if you don't want the openAI support
+    # the one on https://hub.docker.com/r/ugleiton/mongo-gui has a bug
+    # https://github.com/arunbandari/mongo-gui/issues/68
+    image: mongo-gui
+    ports:
+      - "4321:4321"
+    environment:
+      - MONGO_URL=mongodb://root:dawis@dawis_mongo:27017
+    networks:
+      - dawis
+    restart: unless-stopped
+
+  dawis_mysql_gui:
+    container_name: dawis_mysql_gui
+    image: phpmyadmin
+    ports:
+      - "8090:80"
+    environment:
+      - PMA_HOST=dawis_mysql
+      - PMA_USER=root
+      - PMA_PASSWORD=dawis
+    networks:
+      - dawis
+    restart: unless-stopped
+
+  dawis_redis:
+    container_name: dawis_redis
+    image: redis:7.2
+    networks:
+      - dawis
+    deploy:
+      resources:
+        limits:
+          cpus: "1"
+          memory: 1G
+    command: ["redis-server", "--appendonly", "no", "--maxmemory", "1gb", "--maxmemory-policy", "allkeys-lru"]
+    restart: unless-stopped
+
+networks:
+  dawis:
+    name: dawis
diff --git a/pagespeed_migration.py b/pagespeed_migration.py
new file mode 100644
index 0000000..3aa4ede
--- /dev/null
+++ b/pagespeed_migration.py
@@ -0,0 +1,116 @@
+from google.cloud.bigquery.client import Client
+from google.cloud.bigquery.dataset import DatasetReference
+from google.cloud.bigquery.enums import SqlTypeNames
+from google.cloud.bigquery.job import LoadJobConfig, WriteDisposition
+from google.cloud.bigquery.schema import SchemaField
+from google.cloud.bigquery.table import TimePartitioning, TimePartitioningType
+from google.oauth2 import service_account
+from os.path import abspath
+from typing import Sequence
+
+
+def create_client(project: str, credentials_file: str) -> Client:
+    credentials = service_account.Credentials.from_service_account_file(abspath(credentials_file))
+    return Client(project, credentials)
+
+
+def run_query(query: str, result_function: callable):
+    query_job = bigquery_client.query(query)
+    row_iterator = query_job.result(page_size=25000)
+
+    for page in row_iterator.pages:
+        result_data = []
+
+        for row in page:
+            result_item = {}
+
+            for column, value in row.items():
+                result_item[column] = value
+
+            result_data.append(result_item)
+
+        result_function(result_data)
+
+
+def upload_pagespeed_data(data: Sequence[dict]):
+    job_config = LoadJobConfig()
+    job_config.write_disposition = WriteDisposition.WRITE_APPEND
+    job_config.time_partitioning = TimePartitioning(type_=TimePartitioningType.DAY, field=SqlTypeNames.DATE)
+
+    loading_experience_schema_fields = (
+        SchemaField('cls', SqlTypeNames.INTEGER),
+        SchemaField('clsGood', SqlTypeNames.FLOAT),
+        SchemaField('clsMedium', SqlTypeNames.FLOAT),
+        SchemaField('clsBad', SqlTypeNames.FLOAT),
+        SchemaField('lcp', SqlTypeNames.INTEGER),
+        SchemaField('lcpGood', SqlTypeNames.FLOAT),
+        SchemaField('lcpMedium', SqlTypeNames.FLOAT),
+        SchemaField('lcpBad', SqlTypeNames.FLOAT),
+        SchemaField('fcp', SqlTypeNames.INTEGER),
+        SchemaField('fcpGood', SqlTypeNames.FLOAT),
+        SchemaField('fcpMedium', SqlTypeNames.FLOAT),
+        SchemaField('fcpBad', SqlTypeNames.FLOAT),
+        SchemaField('fid', SqlTypeNames.INTEGER),
+        SchemaField('fidGood', SqlTypeNames.FLOAT),
+        SchemaField('fidMedium', SqlTypeNames.FLOAT),
+        SchemaField('fidBad', SqlTypeNames.FLOAT),
+        SchemaField('inp', SqlTypeNames.FLOAT),
+        SchemaField('inpGood', SqlTypeNames.FLOAT),
+        SchemaField('inpMedium', SqlTypeNames.FLOAT),
+        SchemaField('inpBad', SqlTypeNames.FLOAT),
+    )
+
+    job_config.schema = (
+        SchemaField('url', SqlTypeNames.STRING, 'REQUIRED'),
+        SchemaField('strategy', SqlTypeNames.STRING, 'REQUIRED'),
+        SchemaField('date', SqlTypeNames.DATETIME, 'REQUIRED'),
+        SchemaField('statusCode', SqlTypeNames.INTEGER),
+        SchemaField('cluster', SqlTypeNames.STRING, 'REQUIRED'),
+        SchemaField('labdata', SqlTypeNames.RECORD, 'REQUIRED', fields=(
+            SchemaField('cls', SqlTypeNames.FLOAT),
+            SchemaField('lcp', SqlTypeNames.FLOAT),
+            SchemaField('fcp', SqlTypeNames.FLOAT),
+            SchemaField('tbt', SqlTypeNames.FLOAT),
+            SchemaField('mpfid', SqlTypeNames.FLOAT),
+            SchemaField('ttfb', SqlTypeNames.FLOAT),
+            SchemaField('performanceScore', SqlTypeNames.FLOAT),
+            SchemaField('serverResponseTime', SqlTypeNames.FLOAT),
+            SchemaField('usesTextCompression', SqlTypeNames.FLOAT),
+            SchemaField('usesLongCacheTtl', SqlTypeNames.FLOAT),
+            SchemaField('domSize', SqlTypeNames.FLOAT),
+            SchemaField('offscreenImages', SqlTypeNames.FLOAT),
+            SchemaField('usesOptimizedImages', SqlTypeNames.FLOAT),
+            SchemaField('usesResponsiveImages', SqlTypeNames.FLOAT),
+            SchemaField('renderBlockingResources', SqlTypeNames.FLOAT),
+            SchemaField('bootupTime', SqlTypeNames.FLOAT),
+            SchemaField('mainthreadWorkBreakdown', SqlTypeNames.FLOAT),
+        )),
+        SchemaField(
+            'originLoadingExperience',
+            SqlTypeNames.RECORD,
+            'REQUIRED',
+            fields=loading_experience_schema_fields
+        ),
+        SchemaField('loadingExperience', SqlTypeNames.RECORD, fields=loading_experience_schema_fields)
+    )
+
+    for data_item in data:
+        data_item['date'] = data_item['date'].strftime('%Y-%m-%dT%H:%M:%S.%f')
+
+    load_job = bigquery_client.load_table_from_json(data, target_table_reference, job_config=job_config)
+    load_job.result()
+
+
+project = 'project-id'
+dataset = 'yourdataset'
+source_table = 'old_pagespeed_table'
+target_table = 'new_pagespeed_table'
+
+bigquery_client = create_client(project, './your-credentials-file-path.json')
+
+target_table_reference = DatasetReference(
+    project,
+    dataset
+).table(target_table)
+
+run_query('SELECT * FROM `{dataset}.{table}`'.format(dataset=dataset, table=source_table), upload_pagespeed_data)
diff --git a/run-debug.sh b/run-debug.sh
new file mode 100755
index 0000000..b722c38
--- /dev/null
+++ b/run-debug.sh
@@ -0,0 +1,3 @@
+#!/bin/bash
+
+/usr/local/bin/python3 -u module-debugger.py
diff --git a/run.sh b/run.sh
new file mode 100644
index 0000000..3eaa558
--- /dev/null
+++ b/run.sh
@@ -0,0 +1,39 @@
+#!/bin/bash
+
+usermod -u ${CELERY_UID} dawis
+
+touch config/delete_me_for_restart
+chown dawis:dawis . && chown -R dawis:dawis $(ls -I config)
+
+CELERY_MAX_MEMORY_PER_CHILD=$(($(cat /sys/fs/cgroup/memory.max)/$CELERY_CONCURRENCY/1000))
+echo Running workers with $CELERY_MAX_MEMORY_PER_CHILD kB max memory.
+
+/usr/local/bin/python3 -m celery \
+    -A dawis \
+    worker \
+    --uid=${CELERY_UID} \
+    --time-limit=${CELERY_TIME_LIMIT} \
+    --concurrency=${CELERY_CONCURRENCY} \
+    --autoscale=${CELERY_CONCURRENCY},1 \
+    --max-memory-per-child=${CELERY_MAX_MEMORY_PER_CHILD} \
+    --logfile=${CELERY_LOGFILE_PATH}/worker.log \
+    --loglevel=${CELERY_LOGLEVEL} &
+
+workerPid=$!
+echo Started worker with pid $workerPid.
+
+/usr/local/bin/python3 -m celery \
+    -A dawis \
+    beat \
+    --uid=${CELERY_UID} \
+    --logfile=${CELERY_LOGFILE_PATH}/beat.log \
+    --schedule=/opt/dawis/var/beat-schedules/schedule.db \
+    --max-interval=60 \
+    --loglevel=${CELERY_LOGLEVEL} &
+
+beatPid=$!
+echo Started scheduler with pid $beatPid.
+
+trap "echo Stopping... && kill -2 $beatPid && kill -2 $workerPid && tail --pid=$workerPid -f /dev/null && exit" SIGTERM
+
+while true; do sleep 1; done