Amazon IoT Analytics is no longer available to new customers. Existing customers of Amazon IoT Analytics can continue to use the service as normal.
Learn more
Custom partitions
Amazon IoT Analytics supports data partitioning so you can organize the data in your data store. When you use data partitioning to organize data, you can query on pruned data. This decreases the amount of data scanned per query and improves latency.
You can partition your data according to message data attributes or attributes added through pipeline activities.
To get started, enable data partitioning in a data store.
Specify one or more data partition dimensions and connect your partitioned data store to an Amazon IoT Analytics pipeline.
Then, write queries that leverage the WHERE
clause to optimize performance.
Create a data store (console)
The following procedure shows you how to create a data store with a custom partition.
To create a data store
-
Sign in to the Amazon IoT Analytics console
. -
In the navigation pane, choose Data stores.
-
On the Data stores page, choose Create data store.
-
On the Specify data store details page, enter basic information about your data store.
-
For Data store ID, enter a unique data store ID. You can't change this ID after you create it.
-
(Optional) For Tags, choose Add new tag to add one or more custom tags (key-value pairs) to your data store. Tags can help you identify resources that you create for Amazon IoT Analytics.
-
Choose Next.
-
-
On the Configure storage type page, specify how to store your data.
-
For Storage type, choose Service managed storage.
-
For Configure how long you want to keep your processed data, choose Indefinitely.
-
Choose Next.
-
-
On the Configure data format page, define the structure and format of your data records.
-
For your data store data format Classification, choose JSON or Parquet. For more information about Amazon IoT Analytics supported file types, see File formats.
Note
You can't change this format after you create the data store.
-
Choose Next.
-
-
Create custom partitions for this data store.
-
For Add data partitions, select Enable.
-
For Data partition source, specify basic information about the source of your partition.
Choose Sample source, and select the Amazon IoT Analytics channel that collects messages for this data store.
-
For Message sample attributes, select the message attributes you want to use to partition your data store. Then, add your selections as attribute partition dimensions or timestamp partition dimensions under Actions.
Note
You can add only one timestamp partition to your data store.
-
For Custom data store partition dimensions, define basic information about your partition dimensions. Each message sample attribute you selected in the previous step will become the dimensions of your partition. Customize each dimension with these options:
-
Partition type - Specify if this partition dimension is an Attribute or a Timestamp partition type.
-
Attribute name and Dimension name - By default, Amazon IoT Analytics will use the name of the message sample attribute you selected as an identifier for your attribute partition dimension. Edit the attribute name to customize the name of your partition dimension. You can use the dimension name in the
WHERE
clause to optimize query performance.-
The name of any partition attribute dimension is prefixed with
__partition_
. -
For timestamp partition types, Amazon IoT Analytics creates the following four dimensions with names
__year
,__month
,__day
,__hour
.
-
-
Ordering - Rearrange your partition dimensions to improve the latency for your queries.
For Timestamp format, specify the format of your timestamp partition by matching the ingested timestamp from your message data. You can choose one of Amazon IoT Analytics listed format options, or specify one that matches the format of your data. Learn more about specifying date time formatters
. To add a new dimension that isn’t a message attribute, choose Add new partitions.
-
-
Choose Next.
-
-
On the Review and create page, review your choices, and then choose Create data store.
Important
You can't change the data store ID after you create the data store.
To edit existing partitions, you must create another data store and reprocess the data through a pipeline.
-
Verify that your new data store appears on the Data stores page.