Adjacency list pattern in DynamoDB

Introduction

The adjacency list pattern helps you store and find relationships between items, like family trees or company teams, in a simple way.

You want to represent a family tree where each person has parents and children.

You need to model an organization chart showing who reports to whom.

You want to track categories and subcategories in a product catalog.

You want to store comments and replies in a discussion thread.

You want to find all connected items starting from one item.

Syntax

DynamoDB

Table: Items
Attributes:
  - id (Primary Key)
  - parent_id (can be null or id of another item)
  - other attributes...

To find children of an item:
  Query where parent_id = :id_value

The id uniquely identifies each item.

The parent_id points to the item's parent, or is null if it has none.

Examples

This shows a root item with two children.

DynamoDB

id: 1, parent_id: null, name: 'Root Node'
id: 2, parent_id: 1, name: 'Child A'
id: 3, parent_id: 1, name: 'Child B'

Items can have children and multiple root items can exist.

DynamoDB

id: 4, parent_id: 2, name: 'Grandchild of Root'
id: 5, parent_id: null, name: 'Another Root'

This query finds all children of the item with id 1.

DynamoDB

Query: SELECT * FROM Items WHERE parent_id = 1

This query finds all root items (items without parents).

DynamoDB

Query: SELECT * FROM Items WHERE parent_id IS NULL

Sample Program

This example creates a table, adds a root item and two children, then queries the children of the root.

DynamoDB

/* DynamoDB example using AWS CLI syntax for clarity */

# Create table
aws dynamodb create-table \
    --table-name Items \
    --attribute-definitions AttributeName=id,AttributeType=S \
    --key-schema AttributeName=id,KeyType=HASH \
    --provisioned-throughput ReadCapacityUnits=5,WriteCapacityUnits=5

# Insert root item
aws dynamodb put-item --table-name Items --item '{"id": {"S": "1"}, "parent_id": {"NULL": true}, "name": {"S": "Root Node"}}'

# Insert child items
aws dynamodb put-item --table-name Items --item '{"id": {"S": "2"}, "parent_id": {"S": "1"}, "name": {"S": "Child A"}}'
aws dynamodb put-item --table-name Items --item '{"id": {"S": "3"}, "parent_id": {"S": "1"}, "name": {"S": "Child B"}}'

# Query children of root
aws dynamodb scan --table-name Items --filter-expression "parent_id = :pid" --expression-attribute-values '{":pid": {"S": "1"}}'

Important Notes

The adjacency list pattern is simple but can be slow for deep trees because you query one level at a time.

Time complexity for finding direct children is fast (depends on number of children), but finding all descendants requires multiple queries.

Use this pattern when your relationships are mostly one-level or you don't need to query deep hierarchies often.

Summary

The adjacency list pattern stores each item's parent to represent relationships.

It is easy to understand and works well for simple hierarchies.

Querying deep or complex trees may need extra logic or different patterns.

Practice

1. What does the adjacency list pattern store in each item to represent relationships in DynamoDB?

easy

A. The parent item's identifier

B. The list of all child items

C. The full path from root to the item

D. The depth level of the item in the hierarchy

Solution

1. Step 1: Understand adjacency list pattern basics

   The adjacency list pattern stores a reference to the parent item for each node to represent hierarchy.

2. Step 2: Identify what is stored per item

   Each item contains the parent's ID, not the children or full path or depth.

3. Final Answer:

   The parent item's identifier -> Option A

4. Quick Check:

   Adjacency list = parent ID stored [OK]

Hint: Remember: adjacency list stores parent ID only [OK]

Common Mistakes:

• Thinking it stores all children in each item
• Confusing with storing full path or depth
• Assuming it stores unrelated metadata

2. Which of the following is the correct way to represent a parent-child relationship in a DynamoDB item using the adjacency list pattern?

easy

A. { "ItemId": "123", "Children": ["456", "789"] }

B. { "ItemId": "123", "ParentId": "456" }

C. { "ItemId": "123", "Path": "/root/456/123" }

D. { "ItemId": "123", "Depth": 2 }

Solution

1. Step 1: Identify adjacency list item structure

   Adjacency list stores the parent ID inside each item to link it to its parent.

2. Step 2: Match correct JSON structure

   { "ItemId": "123", "ParentId": "456" } shows an item with its own ID and a ParentId field, which is correct.

3. Final Answer:

   { "ItemId": "123", "ParentId": "456" } -> Option B

4. Quick Check:

   ParentId field present = adjacency list format [OK]

Hint: Look for ParentId key in item JSON [OK]

Common Mistakes:

• Using Children array instead of ParentId
• Including full path or depth instead of parent reference
• Confusing item ID with parent ID

3. Given the following DynamoDB items using adjacency list pattern:

{ "ItemId": "1", "ParentId": null }

{ "ItemId": "2", "ParentId": "1" }

{ "ItemId": "3", "ParentId": "1" }

{ "ItemId": "4", "ParentId": "2" }

If you query for items with ParentId = '1', which items will you get?

medium

A. [{ "ItemId": "2", "ParentId": "1" }, { "ItemId": "3", "ParentId": "1" }]

B. [{ "ItemId": "1", "ParentId": null }]

C. [{ "ItemId": "4", "ParentId": "2" }]

D. [{ "ItemId": "2", "ParentId": "1" }, { "ItemId": "3", "ParentId": "1" }, { "ItemId": "4", "ParentId": "2" }]

Solution

1. Step 1: Understand query condition

   The query filters items where ParentId equals '1'.

2. Step 2: Identify matching items

   Items with ParentId '1' are ItemId '2' and '3'. Item '4' has ParentId '2', so it does not match.

3. Final Answer:

   [{ "ItemId": "2", "ParentId": "1" }, { "ItemId": "3", "ParentId": "1" }] -> Option A

4. Quick Check:

   Filter ParentId='1' returns items 2 and 3 [OK]

Hint: Filter by ParentId to get direct children only [OK]

Common Mistakes:

• Including items with different ParentId values
• Confusing root item with children
• Assuming query returns grandchildren

4. You wrote this DynamoDB query to get children of item '5':

FilterExpression: "ParentId = :pid", ExpressionAttributeValues: { ":pid": "5" }

But it returns no results, even though you know children exist. What is the likely error?

medium

A. The ExpressionAttributeValues syntax is incorrect

B. You should use KeyConditionExpression instead of FilterExpression

C. You need to query with ItemId = '5' instead

D. ParentId attribute is missing or named differently in items

Solution

1. Step 1: Check attribute naming

   If ParentId attribute is missing or named differently, filter finds no matches.

2. Step 2: Verify attribute presence in items

   Ensure all child items have ParentId attribute exactly named 'ParentId'. If not, query returns empty.

3. Final Answer:

   ParentId attribute is missing or named differently in items -> Option D

4. Quick Check:

   Attribute name mismatch causes no results [OK]

Hint: Check attribute names match exactly in query and items [OK]

Common Mistakes:

• Confusing FilterExpression with KeyConditionExpression
• Incorrect syntax in ExpressionAttributeValues
• Querying by ItemId instead of ParentId

5. You want to retrieve the full hierarchy path from a child item up to the root using the adjacency list pattern in DynamoDB. Which approach is best?

hard

A. Use a DynamoDB join operation to combine parent and child items

B. Store the full path string in each item and query it directly

C. Iteratively query each item's ParentId until reaching the root

D. Use a single query filtering all items with ParentId equal to the child's ID

Solution

1. Step 1: Understand adjacency list limitations

   Adjacency list stores only parent references, so full path is not stored.

2. Step 2: Retrieve full path by iterative queries

   To get full path, query each item's parent repeatedly until root (ParentId null) is reached.

3. Step 3: Evaluate other options

   Store the full path string in each item and query it directly is a different pattern (materialized path). Use a single query filtering all items with ParentId equal to the child's ID gets children, not parents. Use a DynamoDB join operation to combine parent and child items is invalid as DynamoDB has no join.

4. Final Answer:

   Iteratively query each item's ParentId until reaching the root -> Option C

5. Quick Check:

   Adjacency list needs iterative parent queries for full path [OK]

Hint: Follow parent links step-by-step to build full path [OK]

Common Mistakes:

• Expecting DynamoDB to support joins
• Assuming full path is stored in adjacency list
• Trying to get full path with single query filtering children