> ## Documentation Index
> Fetch the complete documentation index at: https://www.paradedb.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Go to Production

> Move from the sandbox to a production deployment

The Getting Started flow shows that ParadeDB can search, retrieve vectors,
rank, filter, and aggregate from inside Postgres. A production rollout is
mostly about choosing the right topology, indexing the right fields, and giving
Postgres enough room to do the work.

## Production Checklist

1. Choose the deployment model that matches your primary database.
2. Load representative data, not only the small `mock_items` table.
3. Design indexes around real query patterns.
4. Tune index creation, reads, and writes with production-sized data.
5. Confirm query plans with `EXPLAIN`.
6. Decide how upgrades, backups, and high availability will work.

## Production Readiness

[ParadeDB Community](https://github.com/paradedb/paradedb), our open-source
edition, has been downloaded millions of times. It is durable and
crash-safe: changes to ParadeDB indexes are write-ahead logged and support crash
recovery and point-in-time recovery. See [Guarantees](/docs/concepts/guarantees) for
the full details.

[ParadeDB Enterprise](/docs/operate/deploy/enterprise) adds commercial licensing,
support, and physical replication for ParadeDB indexes. Physical replication is
what enables high availability, failover, and read replicas that can serve
ParadeDB queries.

## Deployment

<CardGroup cols={2}>
  <Card title="Deployment Overview" icon="server" href="/docs/operate/deploy/overview">
    Compare extension, self-hosted, logical replication, cloud platform, BYOC,
    and Enterprise options.
  </Card>

  <Card title="Install as an Extension" icon="database" href="/docs/operate/deploy/self-hosted/extension">
    Add ParadeDB to a Postgres instance you already control.
  </Card>

  <Card title="Logical Replication" icon="arrows-rotate" href="/docs/operate/deploy/logical-replication/getting-started">
    Replicate from managed Postgres when the primary cannot run ParadeDB.
  </Card>

  <Card title="High Availability" icon="shield-check" href="/docs/operate/deploy/self-hosted/high-availability">
    Plan read replicas, failover behavior, and production availability.
  </Card>
</CardGroup>

## Index Design

Start from the queries your application actually runs. Add fields to the
ParadeDB index when they need to participate in full-text search, vector
retrieval, filtering, `ORDER BY`, `GROUP BY`, or aggregations.

<CardGroup cols={2}>
  <Card title="Create an Index" icon="list-tree" href="/docs/reference/indexing/create-index">
    Index table columns, tokenizer casts, and framework-native schema examples.
  </Card>

  <Card title="Choosing a Key Field" icon="key" href="/docs/reference/indexing/create-index#choosing-a-key-field">
    Pick the unique row identifier that ParadeDB stores in the index.
  </Card>

  <Card title="Vector Fields" icon="chart-scatter" href="/docs/reference/indexing/indexing-vectors">
    Index embeddings alongside text and filters for nearest-neighbor and hybrid
    search.
  </Card>

  <Card title="Columnar Fields" icon="columns-3" href="/docs/reference/indexing/columnar">
    Make scalar and tokenizer-backed fields available through columnar storage.
  </Card>

  <Card title="Performance Tuning" icon="gauge" href="/docs/operate/performance-tuning/overview">
    Tune index creation, read queries, and write-heavy workloads.
  </Card>
</CardGroup>

## Operational Readiness

Before cutting over traffic, verify the important paths:

* Run representative queries with `EXPLAIN` and confirm ParadeDB is used.
* Test writes and updates against indexed tables.
* Check backup, restore, and upgrade procedures.
* Set monitoring around slow queries, index size, and segment growth.

When something does not look right, compare the query plan with the relevant
query reference page and the [Read Throughput](/docs/operate/performance-tuning/reads)
notes.