Step 5: Create a DynamoDB data model
Define the partition keys for your base table and global secondary indexes (GSIs):
-
Following the key design best practices, use
ComponentIdas the partition key for the base table in this example. Because it's unique,ComponentIdcan offer granularity. DynamoDB uses the hash value of your partition key to determine the partition where the data is stored physically. The unique component ID generates a different hash value, which can facilitate distribution of data inside the table. You can query the base table by using aComponentIdpartition key. -
To find immediate children of a component, create a GSI where
ParentIdis the partition key, andComponentIdis the sort key. You can query this GSI by usingParentIdas the partition key. -
To find all recursive children of a component, create a GSI where
GraphIdis the partition key, andPathis the sort key. You can query this GSI by usingGraphIdas the partition key and theBEGINS_WITH(Path, "$path")operator on the sort key.
Partition key |
Sort Key |
Mapping attributes |
|
Base table |
|
|
|
GSI1 |
|
|
|
GSI2 |
|
|
|
Storing components in the table
The next step is to store each component in the DynamoDB base table. After you insert all the components from the example tree, you get the following base table.
ComponentId |
ParentId |
GraphId |
Path |
CM1 |
CM1#1 |
CM1 |
|
CM2 |
CM1 |
CM1#1 |
CM1|CM2 |
CM3 |
CM1 |
CM1#1 |
CM1|CM3 |
CM4 |
CM2 |
CM1#1 |
CM1|CM2|CM4 |
CM5 |
CM2 |
CM1#1 |
CM1|CM2|CM5 |
CM6 |
CM3 |
CM1#1 |
CM1|CM3|CM6 |
CM7 |
CM3 |
CM1#1 |
CM1|CM3|CM7 |
CM8 |
CM4 |
CM1#1 |
CM1|CM2|CM4|CM8 |
CM9 |
CM4 |
CM1#1 |
CM1|CM2|CM4|CM9 |
CM10 |
CM5 |
CM1#1 |
CM1|CM2|CM5|CM10 |
The GSI1 index
To check all immediate children of a component, you create an index that uses
ParentId as a partition key and ComponentId as a sort key. The
following pivot table represents the GSI1 index. You can use this index to retrieve all
immediate child components by using a parent component ID. For example, you can find out
how many batteries are available in a car (CM1) or which cells are available in a module
(CM4).
ParentId |
ComponentId |
CM1 |
CM2 CM3 |
CM2 |
CM4 CM5 |
CM3 |
CM6 CM7 |
CM4 |
CM8 CM9 |
CM5 |
CM10 |
The GSI2 index
The following pivot table represents the GSI2 index. It's configured using
GraphId as a partition key and Path as a sort key. Using
GraphId and the begins_with operation on the sort key
(Path), you can find the full lineage of a component in a tree.
GraphId |
Path |
ComponentId |
CM1#1 |
CM1 CM1|CM2 CM1|CM3 CM1|CM2|CM4 CM1|CM2|CM5 CM1|CM2|CM4|CM8 CM1|CM2|CM4|CM9 CM1|CM2|CM5|CM10 CM1|CM3|CM6 CM1|CM3|CM7 |
CM1 CM2 CM3 CM4 CM5 CM8 CM9 CM10 CM6 CM7 |
