# Quickstart

<!-- -->

Get TAG running in under 5 minutes. TAG caches your Tigris objects on local disk so repeated reads are served in microseconds instead of milliseconds — and you don't need to change any application code.

## Prerequisites[​](#prerequisites "Direct link to Prerequisites")

You need a pair of Tigris credentials (`AWS_ACCESS_KEY_ID` / `AWS_SECRET_ACCESS_KEY`) with read-only access to all buckets in your Tigris organization. These are TAG's service credentials. Your application clients authenticate separately with their own credentials.

## Install and run[​](#install-and-run "Direct link to Install and run")

* Native Binary
* Docker
* Kubernetes

```
# Download and install TAG

curl -sSL https://raw.githubusercontent.com/tigrisdata/tag-deploy/main/native/install.sh | bash



# Set your Tigris credentials

export AWS_ACCESS_KEY_ID=<your-access-key>

export AWS_SECRET_ACCESS_KEY=<your-secret-key>



# Start TAG

tag --config /etc/tag/config.yaml
```

The install script auto-detects your OS (Linux/macOS) and architecture (amd64/arm64) and places the binary in `/usr/local/bin`.

```
git clone https://github.com/tigrisdata/tag-deploy.git

cd tag-deploy



# Create docker/.env with credentials

cat > docker/.env <<EOF

AWS_ACCESS_KEY_ID=your_access_key

AWS_SECRET_ACCESS_KEY=your_secret_key

EOF



# Start TAG

cd docker && docker compose up -d
```

The Docker image (`tigrisdata/tag`) uses Alpine Linux, runs as a non-root user, and persists cache data in a named volume.

For cluster mode and more options, see the [Docker guide](/docs/acceleration-gateway/docker/.md).

```
git clone https://github.com/tigrisdata/tag-deploy.git

cd tag-deploy



kubectl create namespace tag



kubectl create secret generic tag-credentials \

  --namespace tag \

  --from-literal=AWS_ACCESS_KEY_ID=your_access_key \

  --from-literal=AWS_SECRET_ACCESS_KEY=your_secret_key



kubectl apply -k kubernetes/base/ -n tag
```

This deploys a 3-replica StatefulSet with embedded caching, gossip-based cluster discovery, and a LoadBalancer service on port 8080.

For production considerations and scaling, see the [Kubernetes guide](/docs/acceleration-gateway/kubernetes/.md).

## Verify it works[​](#verify-it-works "Direct link to Verify it works")

TAG is now listening on `http://localhost:8080`. Try a few requests to make sure everything is connected:

```
# Health check

curl http://localhost:8080/health



# List buckets

aws s3 ls --endpoint-url http://localhost:8080



# Download an object

aws s3 cp s3://my-bucket/my-key ./local-file --endpoint-url http://localhost:8080
```

## Connect your application[​](#connect-your-application "Direct link to Connect your application")

TAG works with any S3 client. Just change your endpoint URL to point at TAG and make sure path-style addressing is enabled — that's it.

Path-Style Addressing Only

TAG supports path-style addressing (`http://host:port/bucket/key`). Virtual-hosted style (`http://bucket.host:port/key`) is not supported. Ensure your S3 client is configured for path-style access.

For client examples covering AWS CLI, Python boto3, streaming large files, and troubleshooting, see [S3 Client Usage](/docs/acceleration-gateway/usage/.md).

## Troubleshooting[​](#troubleshooting "Direct link to Troubleshooting")

If something isn't working — TAG won't start, you're getting 403s, or every response shows `X-Cache: MISS` — see [Troubleshooting](/docs/acceleration-gateway/deployment-guide/.md#troubleshooting) in the Deployment Guide.