DynamoDB: Secondary Indexes and Querying

Before building performant applications with DynamoDB, we need to understand how data is accessed and retrieved efficiently. This lesson focused on the fundamental operations and indexing strategies that shape how we interact with data at scale.

Query vs. scan

DynamoDB provides two primary operations to retrieve multiple items:

• Query operation: A query is used when we know the partition key and want to retrieve all (or a subset of) items that share that key. Queries can be efficient, as DynamoDB uses the partition key to locate the appropriate data partitions and then applies optional filters to narrow the results further.

• Scan operation: A scan reads every item in the table, regardless of key. This makes scans significantly more resource-intensive, especially as the table size grows. Scans are often used during data migrations, audits, or administrative tasks, not during routine application queries.

The illustration below depicts the impact of a scan on the provisioned throughput of a DynamoDB table. Notice that as the size of requests becomes nonuniform, throttling (shown by the rectangle bar) becomes more frequent.

Here are some of the strategies to minimize the impact of a scan operation:

• Reduce page size using the Limit parameter: Setting a smaller page size allows scans to consume fewer read capacity units per request. This spreads out the load and ...