# CREATE MATERIALIZED VIEW `CREATE MATERIALIZED VIEW` defines a view that is persisted in durable storage and incrementally updated as new data arrives. Use `CREATE MATERIALIZED VIEW` to: - Create a materialized view that maintains [fresh results](/concepts/reaction-time) by persisting them in durable storage and incrementally updating them as new data arrives. - Create a replacement for an existing materialized view that can be applied in place with [`ALTER MATERIALIZED VIEW ... APPLY REPLACEMENT`](/sql/alter-materialized-view/). Materialized views are particularly useful when you need **cross-cluster access** to results or want to sink data to external systems like [Kafka](/sql/create-sink). When you create a materialized view, a [cluster](/concepts/clusters/), responsible for maintaining the view, is associated with it, but the results can be **queried from any cluster**. This allows you to separate the compute resources used for view maintenance from those used for serving queries. If you do not need durability or cross-cluster sharing, and you are primarily interested in fast query performance within a single cluster, you may prefer to [create a view and index it](/concepts/views/#views). In Materialize, [indexes on views](/concepts/indexes/) also maintain results incrementally, but store them in memory, scoped to the cluster where the index was created. This approach offers lower latency for direct querying within that cluster. ## Syntax **CREATE MATERIALIZED VIEW:** ### Create materialized view ```mzsql CREATE MATERIALIZED VIEW [IF NOT EXISTS] [(, ...)] [IN CLUSTER ] [WITH ()] AS ; ``` | Syntax element | Description | | --- | --- | | `IF NOT EXISTS` | If specified, do not generate an error if a materialized view of the same name already exists. | | `` | A name for the materialized view. | | `(, ...)` | Rename the `SELECT` statement's columns to the list of identifiers. Both must be the same length. Note that this is required for statements that return multiple columns with the same identifier. | | `IN CLUSTER ` | The cluster to maintain this materialized view. If not specified, defaults to the active cluster. | | `WITH ()` | The following `` are supported: \| Field \| Value \| Description \| \|-------\|-------\|-------------\| \| `ASSERT NOT NULL` *col_ident* \| `text` \| The column identifier for which to create a [non-null assertion](#non-null-assertions). To specify multiple columns, use the option multiple times. \| \| `PARTITION BY` *columns* \| `(ident [, ident]*)` \| The key by which Materialize should internally partition this durable collection. See the [partitioning guide](/transform-data/patterns/partition-by/) for restrictions on valid values and other details. \| \| `RETAIN HISTORY FOR` *retention_period* \| `interval` \| ***Private preview.*** Duration for which Materialize retains historical data, which is useful to implement [durable subscriptions](/transform-data/patterns/durable-subscriptions/#history-retention-period). Accepts positive [interval](/sql/types/interval/) values (e.g. `'1hr'`). Default: `1s`. \| \| `REFRESH` *refresh_strategy* \| \| ***Private preview.*** The refresh strategy for the materialized view. See [Refresh strategies](#refresh-strategies) for syntax options. Default: `ON COMMIT`. \| | | `` | The [`SELECT` statement](/sql/select) whose results you want to maintain incrementally updated. | **CREATE REPLACEMENT MATERIALIZED VIEW:** ### Create replacement materialized view > **Public Preview:** This feature is in public preview. Create a replacement materialized view for an existing materialized view. ```mzsql CREATE REPLACEMENT MATERIALIZED VIEW FOR [IN CLUSTER ] [WITH ()] AS ; ``` | Syntax element | Description | | --- | --- | | `` | A name for the replacement materialized view. | | `` | The name of the existing materialized view to be replaced. The replacement materialized view can only be applied to this materialized view. | | `IN CLUSTER ` | The cluster to maintain this replacement materialized view. If not specified, defaults to the active cluster. | | `WITH ()` | Same options as `CREATE MATERIALIZED VIEW`. | | `` | The [`SELECT` statement](/sql/select) for the replacement view. The statement must produce the same output schema as the target materialized view; i.e., column names, column types, column order, nullability, and keys must all match. | The created replacement materialized view starts hydrating immediately and can later be applied to replace the specified materialized view. For more information, see [Creating replacement materialized views](#creating-replacement-materialized-views). ## Details ### Usage pattern In Materialize, both indexes on views and materialized views incrementally update the view results when Materialize ingests new data. Whereas materialized views persist the view results in durable storage and can be accessed across clusters, indexes on views compute and store view results in memory within a single cluster.

Some general guidelines for usage patterns include:

Usage Pattern General Guideline
View results are accessed from a single cluster only;
such as in a 1-cluster or a 2-cluster architecture.		 View with an index
View used as a building block for stacked views; i.e., views not used to serve results. View
View results are accessed across clusters;
such as in a 3-cluster architecture.		 Materialized view (in the transform cluster)
Index on the materialized view (in the serving cluster)
Use with a sink or a SUBSCRIBE operation Materialized view
Use with temporal filters Materialized view
### Indexing materialized views Although you can query a materialized view directly, these queries will be issued against Materialize's storage layer. This is expected to be fast, but still slower than reading from memory. To improve the speed of queries on materialized views, we recommend creating [indexes](../create-index) based on common query patterns. It's important to keep in mind that indexes are **local** to a cluster, and maintained in memory. As an example, if you create a materialized view and build an index on it in the `quickstart` cluster, querying the view from a different cluster will _not_ use the index; you should create the appropriate indexes in each cluster you are referencing the materialized view in. [//]: # "TODO(morsapaes) Point to relevant operational guide on indexes once this exists+add detail about using indexes to optimize materialized view stacking." ### Non-null assertions Because materialized views may be created on arbitrary queries, it may not in all cases be possible for Materialize to automatically infer non-nullability of some columns that can in fact never be null. In such a case, `ASSERT NOT NULL` clauses may be used as described in the syntax section above. Specifying `ASSERT NOT NULL` for a column forces that column's type in the materialized view to include `NOT NULL`. If this clause is used erroneously, and a `NULL` value is in fact produced in a column for which `ASSERT NOT NULL` was specified, querying the materialized view will produce an error until the offending row is deleted. ### Refresh strategies Materialized views in Materialize are incrementally maintained by default, meaning their results are automatically updated as soon as new data arrives. This guarantees that queries returns the most up-to-date information available with minimal delay and that results are always as [fresh](/concepts/reaction-time) as the input data itself. In most cases, this default behavior is ideal. However, in some very specific scenarios like reporting over slow changing historical data, it may be acceptable to relax freshness in order to reduce compute usage. For these cases, Materialize supports refresh strategies, which allow you to configure a materialized view to recompute itself on a fixed schedule rather than maintaining them incrementally. > **Note:** The use of refresh strategies is discouraged unless you have a clear and measurable need to reduce maintenance costs on stale or archival data. For most use cases, the default incremental maintenance model provides a better experience. [//]: # "TODO(morsapaes) We should add a SQL pattern that walks through a full-blown example of how to implement the cold, warm, hot path with refresh strategies." #### Refresh on commit

Syntax: REFRESH ON COMMIT

Materialized views in Materialize are incrementally updated by default. This means that as soon as new data arrives in the system, any dependent materialized views are automatically and continuously updated. This behavior, known as **refresh on commit**, ensures that the view's contents are always as fresh as the underlying data. **`REFRESH ON COMMIT` is:** * **Generally available** * The **default behavior** for all materialized views * **Implicit** and does not need to be manually specified * **Strongly recommended** for the vast majority of use cases With `REFRESH ON COMMIT`, Materialize provides low-latency, up-to-date results without requiring user-defined schedules or manual refreshes. This model is ideal for most workloads, including streaming analytics, live dashboards, customer-facing queries, and applications that rely on timely, accurate results. Only in rare cases—such as batch-oriented processing or reporting over slowly changing historical datasets—might it make sense to trade off freshness for potential cost savings. In such cases, consider defining an explicit [refresh strategy](#refresh-strategies) to control when recomputation occurs. #### Refresh at

Syntax: REFRESH AT { CREATION | timestamp }

This strategy allows configuring a materialized view to **refresh at a specific time**. The refresh time can be specified as a timestamp, or using the `AT CREATION` clause, which triggers a first refresh when the materialized view is created. **Example** To create a materialized view that is refreshed at creation, and then at the specified times: ```mzsql CREATE MATERIALIZED VIEW mv_refresh_at IN CLUSTER my_scheduled_cluster WITH ( -- Refresh at creation, so the view is populated ahead of -- the first user-specified refresh time REFRESH AT CREATION, -- Refresh at a user-specified (future) time REFRESH AT '2024-06-06 12:00:00', -- Refresh at another user-specified (future) time REFRESH AT '2024-06-08 22:00:00' ) AS SELECT ... FROM ...; ``` You can specify multiple `REFRESH AT` strategies in the same `CREATE` statement, and combine them with the [`REFRESH EVERY` strategy](#refresh-every). #### Refresh every

Syntax: REFRESH EVERY interval [ ALIGNED TO timestamp ]

This strategy allows configuring a materialized view to **refresh at regular intervals**. The `ALIGNED TO` clause additionally allows specifying the _phase_ of the scheduled refreshes: for daily refreshes, it specifies the time of the day when the refresh will happen; for weekly refreshes, it specifies the day of the week and the time of the day when the refresh will happen. If `ALIGNED TO` is not specified, it defaults to the time when the materialized view is created. **Example** To create a materialized view that is refreshed at creation, and then once a day at 10PM UTC: ```mzsql CREATE MATERIALIZED VIEW mv_refresh_every IN CLUSTER my_scheduled_cluster WITH ( -- Refresh at creation, so the view is populated ahead of -- the first user-specified refresh time REFRESH AT CREATION, -- Refresh every day at 10PM UTC REFRESH EVERY '1 day' ALIGNED TO '2024-06-06 22:00:00' ) AS SELECT ...; ``` You can specify multiple `REFRESH EVERY` strategies in the same `CREATE` statement, and combine them with the [`REFRESH AT` strategy](#refresh-at). When this strategy, we recommend **always** using the [`REFRESH AT CREATION`](#refresh-at) clause, so the materialized view is available for querying ahead of the first user-specified refresh time. #### Querying materialized views with refresh strategies Materialized views configured with [`REFRESH EVERY` strategies](#refresh-every) have a period of unavailability around the scheduled refresh times — during this period, the view **will not return any results**. To avoid unavailability during the refresh operation, you must host these views in [**scheduled clusters**](/sql/create-cluster/#scheduling), which can be configured to automatically [turn on ahead of the scheduled refresh time](/sql/create-cluster/#hydration-time-estimate). **Example** To create a scheduled cluster that turns on 1 hour ahead of any scheduled refresh times: ```mzsql CREATE CLUSTER my_scheduled_cluster ( SIZE = '3200cc', SCHEDULE = ON REFRESH (HYDRATION TIME ESTIMATE = '1 hour') ); ``` You can then create a materialized view in this cluster, configured to refresh at creation, then once a day at 12PM UTC: ```mzsql CREATE MATERIALIZED VIEW mv_refresh_every IN CLUSTER my_scheduled_cluster WITH ( -- Refresh at creation, so the view is populated ahead of -- the first user-specified refresh time REFRESH AT CREATION, -- Refresh every day at 12PM UTC REFRESH EVERY '1 day' ALIGNED TO '2024-06-18 00:00:00' ) AS SELECT ...; ``` Because the materialized view is hosted on a scheduled cluster that is configured to **turn on ahead of any scheduled refreshes**, you can expect `my_scheduled_cluster` to be provisioned at 11PM UTC — or, 1 hour ahead of the scheduled refresh time for `mv_refresh_every`. This means that the cluster can backfill the view with pre-existing data — a process known as [_hydration_](/transform-data/troubleshooting/#hydrating-upstream-objects) — ahead of the refresh operation, which **reduces the total unavailability window of the view** to just the duration of the refresh. If the cluster is **not** configured to turn on ahead of scheduled refreshes (i.e., using the `HYDRATION TIME ESTIMATE` option), the total unavailability window of the view will be a combination of the hydration time for all objects in the cluster (typically long) and the duration of the refresh for the materialized view (typically short). Depending on the actual time it takes to hydrate the view or set of views in the cluster, you can later adjust the hydration time estimate value for the cluster using [`ALTER CLUSTER`](../alter-cluster/#schedule): ```mzsql ALTER CLUSTER my_scheduled_cluster SET (SCHEDULE = ON REFRESH (HYDRATION TIME ESTIMATE = '30 minutes')); ``` #### Introspection To check details about the (non-default) refresh strategies associated with any materialized view in the system, you can query the [`mz_internal.mz_materialized_view_refresh_strategies`](/reference/system-catalog/mz_internal/#mz_materialized_view_refresh_strategies) and [`mz_internal.mz_materialized_view_refreshes`](/reference/system-catalog/mz_internal/#mz_materialized_view_refreshes) system catalog tables: ```mzsql SELECT mv.id AS materialized_view_id, mv.name AS materialized_view_name, rs.type AS refresh_strategy, rs.interval AS refresh_interval, rs.aligned_to AS refresh_interval_phase, rs.at AS refresh_time, r.last_completed_refresh, r.next_refresh FROM mz_internal.mz_materialized_view_refresh_strategies rs JOIN mz_internal.mz_materialized_view_refreshes r ON r.materialized_view_id = rs.materialized_view_id JOIN mz_materialized_views mv ON rs.materialized_view_id = mv.id; ``` ### Creating replacement materialized views > **Public Preview:** This feature is in public preview. You can use [`CREATE REPLACEMENT MATERIALIZED VIEW`](/sql/create-materialized-view/) with [`ALTER MATERIALIZED VIEW ... APPLY REPLACEMENT`](/sql/alter-materialized-view) to replace materialized views in-place without recreating dependent objects or incurring downtime.

To create a replacement materialized view, you must:

  • Specify the target materialized view.
  • Specify a SELECT statement for the replacement view that produces the same output schema (including column order and keys) as the target view.

Upon creation, the replacement view starts hydrating in the background.

Before applying the replacement view, verify that the replacement view is hydrated to avoid downtime: The replacement view is dropped when you apply the replacement view. For more information on applying the replacement view, including recommendations and CPU/memory considerations, see [`ALTER MATERIALIZED VIEW ... APPLY REPLACEMENT...`](/sql/alter-materialized-view/#replacing-a-materialized-view) See also: - [Replace materialized views](/transform-data/updating-materialized-views/replace-materialized-view/) guide for a step-by-step tutorial. #### Query performance of replacement views You can query a replacement materialized view to validate its results before replacing. However, when queried, replacement materialized views are treated like a [view](/sql/create-view), and the query results are re-computed as part of the query execution. As such, queries against replacement materialized views are slower and more computationally expensive than queries against regular materialized views. #### Restrictions and limitations A replacement materialized view can only be applied to the target materialized view specified in the `FOR` clause of the [`CREATE REPLACEMENT MATERIALIZED VIEW`](/sql/create-materialized-view/) statement. You cannot create dependent objects using [replacement materialized views](/sql/create-materialized-view/#creating-replacement-materialized-views); for example, you cannot create an index on a replacement materialized view or create other views on a replacement materialized view. ## Examples ### Creating a materialized view The following example creates a `winning_bids` materialized view: ```mzsql CREATE MATERIALIZED VIEW winning_bids AS SELECT DISTINCT ON (a.id) b.*, a.item, a.seller FROM auctions AS a JOIN bids AS b ON a.id = b.auction_id WHERE b.bid_time < a.end_time AND mz_now() >= a.end_time ORDER BY a.id, b.amount DESC, b.bid_time, b.buyer; ``` ### Using non-null assertions ```mzsql CREATE MATERIALIZED VIEW users_and_orders WITH ( -- The semantics of a FULL OUTER JOIN guarantee that user_id is not null, -- because one of `users.id` or `orders.user_id` must be not null, but -- Materialize cannot yet automatically infer that fact. ASSERT NOT NULL user_id ) AS SELECT coalesce(users.id, orders.user_id) AS user_id, ... FROM users FULL OUTER JOIN orders ON users.id = orders.user_id ``` ### Using refresh strategies ```mzsql CREATE MATERIALIZED VIEW mv IN CLUSTER my_refresh_cluster WITH ( -- Refresh every Tuesday at 12PM UTC REFRESH EVERY '7 days' ALIGNED TO '2024-06-04 12:00:00', -- Refresh every Thursday at 12PM UTC REFRESH EVERY '7 days' ALIGNED TO '2024-06-06 12:00:00', -- Refresh on creation, so the view is populated ahead of -- the first user-specified refresh time REFRESH AT CREATION ) AS SELECT ... FROM ...; ``` [//]: # "TODO(morsapaes) Add more elaborate examples with \timing that show things like querying materialized views from different clusters, indexed vs. non-indexed, and so on." ### Creating a replacement materialized view > **Public Preview:** This feature is in public preview. The following example creates a replacement materialized view `winning_bids_replacement` for the `winning_bids` materialized view. The replacement view specifies a different filter `mz_now() > a.end_time` than the existing view `mz_now() >= a.end_time`. ```mzsql CREATE REPLACEMENT MATERIALIZED VIEW winning_bids_replacement FOR winning_bids AS SELECT DISTINCT ON (a.id) b.*, a.item, a.seller FROM auctions AS a JOIN bids AS b ON a.id = b.auction_id WHERE b.bid_time < a.end_time AND mz_now() > a.end_time ORDER BY a.id, b.amount DESC, b.bid_time, b.buyer; ``` To replace the existing view with its replacement, see [`ALTER MATERIALIZED VIEW`](../alter-materialized-view). See also: - [Replace materialized views guide ](/transform-data/updating-materialized-views/replace-materialized-view/) ## Privileges The privileges required to execute this statement are: - `CREATE` privileges on the containing schema. - `CREATE` privileges on the containing cluster. - `USAGE` privileges on all types used in the materialized view definition. - `USAGE` privileges on the schemas for the types used in the statement. ## Additional information - Materialized views are not monotonic; that is, materialized views cannot be recognized as append-only. ## Related pages - [`ALTER MATERIALIZED VIEW`](../alter-materialized-view) - [`SHOW MATERIALIZED VIEWS`](../show-materialized-views) - [`SHOW CREATE MATERIALIZED VIEW`](../show-create-materialized-view) - [`DROP MATERIALIZED VIEW`](../drop-materialized-view)