NoSQL · Data models beyond tables

How do secondary indexes work in NoSQL systems?

Mediumsecondary indexesgsiquery patternsUpdated Oct 2026

Short answer

A secondary index maintains an alternate key so you can query by a non-primary attribute. It adds write cost and, in distributed stores, is often eventually consistent and limited in the queries it supports.

The primary key determines the main access path. A secondary index copies selected fields into a new structure keyed by the query attribute. In a distributed system that index may live on different nodes, so a write must update both the table and the index.

Local versus global

Local index: stored with the base partition, so a query may need to contact every partition
Global index: partitioned by the indexed attribute, so writes may cross partitions
Global indexes are usually eventually consistent because the index update is asynchronous
Some systems support only eventual consistency for secondary index reads

json

{
  "TableName": "Orders",
  "GlobalSecondaryIndexes": [
    {
      "IndexName": "orders_by_status",
      "KeySchema": [
        { "AttributeName": "status", "KeyType": "HASH" },
        { "AttributeName": "createdAt", "KeyType": "RANGE" }
      ],
      "Projection": { "ProjectionType": "INCLUDE", "NonKeyAttributes": ["total"] }
    }
  ]
}

A global secondary index gives orders a second query path by status and date. It has its own capacity and write cost.

Design indexes for real query patterns and project only needed attributes. A common mistake is indexing a low-cardinality field such as status, which creates a large, hot index partition. Another is expecting the index to support filters it was not built for.

Secondary indexes are not free storage. They consume write capacity, disk and sometimes memory, and they can fail to keep up under heavy write load. Monitor index lag and throttling separately from the base table.

Follow-ups you should expect

Interviewers use these to check whether the first answer was understanding or memorisation.

1What is the difference between a local and a global secondary index?
2Why are global secondary indexes often eventually consistent?
3How do you avoid a hot secondary index?
4What does index projection mean?

Say it in the interview

A secondary index creates an alternate query path by copying selected fields into a new key structure. In a distributed store, that means writes update both the table and the index, and the index may be eventually consistent. I choose between local and global based on whether I want the index colocated with the data or partitioned by the indexed field. I avoid indexing low-cardinality fields, because that creates a large hot index partition, and I project only the attributes the query needs.

Use this as a model, not a script. Saying it in your own words is what makes it survive follow-up pressure.

Now say it out loud.

Understanding an answer and delivering it under pressure are different skills. Rehearse this topic in a live NoSQL mock interview and get scored feedback on depth, clarity and confidence.

Start a free NoSQL interview Browse all questions

Local versus global

Follow-ups you should expect

Related questions

Now say it out loud.