🤖 Meet OnCall AI, our observability copilot that makes troubleshooting easy. Read announcement.

Skip to content

What is ElasticSearch: 6 Performance and Cost Optimization Tips

Apr 22, 2024 / 9 minute read

Elasticsearch can be expensive, especially when you scale your clusters. Discover 6 effective ways to optimize Elasticsearch's performance and reduce its costs!


Over 60,000 companies use Elasticsearch to ingest and analyze data in near-real time. Despite the benefits, many users incur increased costs, especially when running Elasticsearch in larger clusters. Fortunately, there's a way to cut its costs and improve its performance: Elasticsearch Optimization.

Optimizing Elasticsearch means implementing several practices to improve efficiency in using this tool. This process helps users understand what resources they use and what they don't. As a result, it reduces costs associated with Elasticsearch while improving its performance.

As more companies use Elasticsearch for data searching, analysis, and management, understanding how to optimize it becomes crucial. This article covers the benefits and better alternatives to reduce or optimize Elasticsearch's cost and performance.

🔑 Key Takeaways

  • Elasticsearch is an effective tool for near-real-time data search, ingestion, and analysis. As a former open-source solution, it now incurs charges, mainly when used as clusters with a subscription.

  • Many companies that rely on Elasticsearch functions encounter high costs as they scale. The expense surge is mainly due to its connection to the Elastic Stack subscription.

  • Besides the expensive nature, the lack of understanding of Elasticsearch also leads to wasted resources. These problems lead to the need for performance and cost optimization.

  • Optimizing Elasticsearch’s performance and cost involves several practices to understand how to maximize resources to reduce costs without sacrificing performance.

  • Besides implementing best practices, users can reduce Elasticsearch costs by adopting more cost-effective alternatives like Edge Delta.

6 Practices to Optimize Elasticsearch Performance and Cost

Despite being an efficient tool, Elasticsearch can cost a lot if not understood and used correctly. Without proper optimization, you may incur high costs for the following reasons:

  • Scaling: Since Elasticsearch clusters are intertwined with subscriptions, they can incur additional costs as you scale.

  • Updates: Updates are crucial for maintaining Elasticsearch systems. However, they are expensive, especially for high-frequency clusters.

  • Experimentation: Since predictions are unreliable, you'll need to experiment with your clusters, which leads to more resource usage.

  • Improper Use of Storage Tiers: It's always tempting to prioritize better performance over costs, so high-level storage seems the best option. However, this also leads to a waste of storage for rarely queried data.

As more companies rely on Elasticsearch, users are becoming aware of the importance of optimizing its costs. Here are some practices and strategies to consider to improve Elasticsearch's performance and reduce its costs.

1. Reduce the Indexed Data Volume

A high volume of indexed data can take up too much storage and slow your Elasticsearch's cluster performance. Reducing it will help reduce storage costs and boost cluster efficiency.

Automating Your Index Lifecycles

With this strategy, you can ensure that every piece of data you index is routed to its ideal storage type. It will also automate data removal based on its usage pattern and age.

Various storage tiers are available for managing storage based on the necessity of each data. These tiers are as follows:

Storage Tier Nodes Features Used for
Hot 6 Nodes Fastest Hardware Profile With Replicas Recently Indexed and Most Searched Data
Warm 4 Nodes Fast Hardware Profile With Replicas Less Searched Data
Cold 2 Nodes Average Hardware Profile Without Replicas Data That Won't Get Any Updates
Frozen 1 Node Slow Hardware Profile Without Replicas Data That's Rarely Queried or Not Queried Anymore

Implementing this strategy requires Index Lifecycle Management policies. These policies allow you to set actions like shrink, delete, rollover, or force merge:

Use Data Tiers and Searchable Snapshots

Besides creating ILM policies, you can also automate data transitions between storage tiers. This practice will minimize storage costs without compromising data accessibility. With these solutions, you can leverage low-cost and long-term storage services, reducing Elasticsearch cluster resources.

âś… Pro-Tip

Use searchable snapshots rather than frozen storage for rarely accessed data. Low-cost object storage providers from big companies like AWS, Azure, and Google can store these snapshots. You can also use Amazon Glacier to archive data for compliance purposes.

Summarizing Older Data

Using Elasticsearch's Rollup API, you can summarize historical data for analysis. After summarizing the historical data, you can delete or archive it. This strategy drastically reduces storage space while maintaining insights. It's an efficient way to deal with voluminous time-series data.

2. Configuring Shard Replicas

Shard replicas are copies of shards but stored in a different node. While replicas ensure data accessibility during node failures, they can still affect cost and performance. With proper configuration, you can optimize these replicas to reduce costs and boost performance.

Adjusting Replicas Depending on Usage

The number of replicas you have should depend on the necessity of the workload. Here's a simple guide you can follow:

  • Read-Heavy Workloads: Increase replica count to improve query performance at the cost of higher storage expense.

  • Writing Operations: Reduce replica count to lower storage costs without compromising writing efficiency.

Leverage Adaptive Replica Selection

Elasticsearch offers Adaptive Replica Selection or ARS. With this feature, you can proactively direct queries to the most suitable replica. This strategy reduces storage and infrastructure costs while improving query performance.

3. Leverage Shard Configurations

Shards are instances of a Lucene index, which is crucial for handling queries about a data subset in a cluster. Plenty of shards promote fast indexing, while fewer enable fast searching. The key is to find the right balance between shard count and workload. This will allow you to reduce costs without compromising performance.

Here are ways to leverage shard configurations:

  • Decide How Many Primary Shards to Use

    Calculate the optimal number of primary shards based on the expected size of your indices. Too many primary shards can increase overhead and cost, while too few can reduce performance. In most cases, 10-50 GB shards are the sweet spot.

  • Filter and Route Queries for Shards

    Elasticsearch offers routing and filtering functions that direct queries to suitable shards. With this strategy, you can reduce the required resources when executing queries. It also lowers the need for extra shards, reducing overall costs.

4. Optimizing Index Templates and Mappings

With proper use of index templates and mappings, you can significantly reduce data to lower storage costs. You can reduce the required fields by using proper field types to optimize your index mapping. It's wise to avoid source field storage as much as possible. This way, you only take a little storage space.

Use a Remove Processor or an Index Pipeline

Elasticsearch allows you to set up a 'remove processor' or an 'index pipeline.' These features allow you to filter out and remove unnecessary and unused fields from your data source.

5. Optimizing Cache Usage and Query Performance

Cache and queries in Elasticsearch complement each other in process and resource usage. By optimizing these two, you can reduce the required resources when handling queries.

Use Query Caching

Query caching is an automated process that stores frequently used queries in a cache. You can adjust the cache size and expiration time to ensure the most relevant data is cached. This step will speed up queries and reduce the load on your cluster.

Leverage Query Optimization Techniques

If you're encountering slow queries, you can optimize them using proven techniques such as the following:

  • Pagination

  • Query Filtering

  • Result Trimming

Elasticsearch also offers an 'explain' API that can detect slow queries. With this function, you can see which queries need adjustments and boost their execution for better efficiency.

Avoid High-Processing Actions

Elasticsearch isn't designed to work well with actions requiring high processing power as a document-oriented data storage engine. Thus, as much as possible, avoid actions like the following on irrelevant data:

  • Parent-type queries

  • Nested queries

  • Nested aggregations

  • Running aggregations

Avoid complex document-index relationships to reduce Elasticsearch costs and boost performance. The most effective approach is to denormalize your data, even if it means storing redundant data in several documents.

6. Downsizing Subscription Costs

If you're using Elasticsearch with Elastic Stack, you'll notice that subscriptions are expensive. The costs are also tied to your cluster's number of nodes. As a result, the costs continue to rise as you scale and create more clusters or increase your nodes.

While running an Elasticsearch cluster using a subscription seems convenient, it's not cost-effective in the long run. If you need the same features that Elasticsearch offers, you can use alternative tools like Edge Delta for lower costs.

Edge Delta: A Cost-Effective Alternative to Elasticsearch

Edge Delta is an observability platform that offers a more cost-effective solution for managing and storing large volumes of data for analysis. It's an ideal alternative if you want a tool compatible with Elastic and other popular tools and platforms.

Here are some of the features Edge Delta offers:

Store and Search Petabytes of Data

Edge Delta allows you to store all your data in low-cost object storage, eliminating the need to drop or sample data. It also has a highly efficient platform that allows quick and efficient querying of vast data.

Direct data to One or More Streaming Destinations

With Edge Delta's vendor-agnostic data routing, you can set up a multi-vendor tiered logging strategy, routing data to the best destination based on it's use case. You can connect Edge Delta with your existing tools for easy machine data analytics and insights with over 60 pre-built integrations, including the Elastic Stack.

Automatic Anomaly Detection

Edge Delta also runs AI/ML to detect anomalous spikes in negative sentiment logs. In doing so, you can alert on issues without configuring monitors. This helps you give time back to your team and reduce false negatives.

Edge Delta's complete set of efficient observability functions helps reduce costs by up to 60%. Moreover, it offers better control over any data type and volume. As an observability platform, Edge Delta offers more than full-text search engine solutions like Elasticsearch.


Elasticsearch offers a viable solution to many companies' need for responsive search and analytics solutions. However, its connection to subscriptions increases costs when companies scale their operations and workloads.

For this reason, it's crucial to understand how to implement specific strategies to optimize its functions and make the most of this tool. Users can maximize resources like storage to boost performance and reduce costs by understanding its structure.

Besides these practices, tools like Edge Delta offer a more cost-effective alternative to Elasticsearch. Its features provide companies with better ways to manage and analyze data without worrying about costs, even when scaling.

FAQS on Elastic Search Optimization

What are the factors that make Elasticsearch expensive?

Elasticsearch becomes more expensive as it scales in infrastructure, maintenance, management, and subscription. Though open-source, it charges fees when you begin running Elasticsearch clusters.

What are the expensive queries in Elasticsearch?

Expensive queries in Elasticsearch use excessive CPU/Memory space. They can cause issues like timeouts and HTTP service errors if not optimized.

Is Elasticsearch free for commercial use?

The free basic Elasticsearch is usable for commercial use. However, this is only true if you don't expose Elasticsearch or Kibana to users directly.


Riley Peronto

Director of Product Marketing

Filed Under


Related Posts

Stay in Touch

Sign up for our newsletter to be the first to know about new articles.