What is Replica in Elasticsearch: Explanation and Examples

In Elasticsearch, a replica is a copy of a primary shard that provides data redundancy and improves search performance. Replicas help protect your data from hardware failures and allow Elasticsearch to handle more search requests by distributing the load.

How It Works

Imagine you have important documents stored in a filing cabinet. To keep them safe, you make copies and store them in another cabinet. In Elasticsearch, the original data is stored in primary shards, and the copies are called replica shards. These replicas are exact copies of the primary shards.

When you add or update data, Elasticsearch writes it to the primary shard first, then copies it to the replicas. This way, if one server or shard fails, the replicas can take over without losing data. Also, replicas help by sharing the work when many users search the data, making searches faster and more reliable.

Example

This example shows how to create an index with one replica in Elasticsearch using a simple JSON request.

json

PUT /my-index
{
  "settings": {
    "number_of_shards": 3,
    "number_of_replicas": 1
  }
}

Output

{
  "acknowledged": true,
  "shards_acknowledged": true,
  "index": "my-index"
}

When to Use

You should use replicas when you want to protect your data from loss and improve search speed. For example, if your Elasticsearch cluster runs on multiple servers, replicas ensure that if one server goes down, your data is still available on another.

Replicas are also useful when your application has many users searching at the same time. By having replicas, Elasticsearch can send search requests to multiple copies, reducing wait times and improving user experience.

Key Points

• Replicas are copies of primary shards that provide fault tolerance.
• They improve search performance by allowing parallel query execution.
• Elasticsearch automatically keeps replicas in sync with primary shards.
• You can configure the number of replicas per index based on your needs.
• Replicas do not store data independently; they depend on primary shards for updates.

Key Takeaways

A replica is a copy of a primary shard that ensures data safety and availability.

Replicas improve search speed by distributing query load across multiple copies.

You can set the number of replicas when creating or updating an index.

Replicas help Elasticsearch recover quickly from hardware or network failures.

Using replicas is essential for production environments needing high reliability.