Elasticsearch Serverless is a managed service that does not require you to manage underlying nodes or resource allocation. It automatically scales resources on demand and is suitable for various scenarios, such as search, log analysis, and vector search. The service offers different application types, including Search-enhanced (v8.17), Log Analysis, and General-purpose Search. These application types have different features and are suited for different scenarios. This topic describes how to select an application type for Elasticsearch Serverless.
Background information
Alibaba Cloud Elasticsearch includes two service types: the cloud-native Platform as a Service (PaaS) version of the Elastic Stack (PaaS version) and Elasticsearch Serverless (Serverless version).
PaaS version:
Supports a wider range of features, versions, and APIs.
Provides greater control over clusters. For example, you can configure the number of nodes and upgrade versions yourself. Node operations, maintenance, and management are more transparent.
Billed based on instance node specifications and storage space. You must estimate resource requirements in advance.
Serverless version:
Offers a managed operations and maintenance (O&M) experience for a simplified user experience. You do not need to manage node deployment or maintenance. The system automatically handles scaling, load balancing, version upgrades, and other operations.
Billed based on the actual consumption of Compute Units (CUs) and storage space. This avoids waste from idle resources.
To ensure better service stability and performance, the Serverless version has some limits compared to the PaaS version. However, the Serverless version has clear advantages in resource scheduling flexibility and O&M automation, focusing on simplifying the developer experience. For a detailed comparison between the Serverless and PaaS versions, see Comparison between ES clusters and Elasticsearch Serverless.
Serverless version selection guide
The Serverless version offers three application types for different scenarios: Search-enhanced (v8.17), Log Analysis, and General-purpose Search. Refer to the following table to select the application type that best suits your business needs.
Application type | Core features | Scenarios | Specification limits and recommendations |
| Suitable for search scenarios with small to medium CU usage. For example, AI search, vector search, and real-time search. | The maximum fixed CU usage for a single application is For information about how to calculate and estimate CU usage, see CU selection reference. | |
Provides built-in, deeply optimized features for log data scenarios such as collection, storage, query, and analysis (for example, hot-cold data separation). It is a highly cost-effective choice for large-scale data ingestion. | Suitable for scenarios involving time series data ingestion, such as logs and metrics. For example, an application log analysis platform. | This application has a minimum consumption threshold of | |
General-purpose Search application (New instances cannot be created) |
| Suitable for general scenarios such as data retrieval and multidimensional analysis. For example, E-commerce product search and user behavior analysis. | Important You can no longer create new instances of this application type. However, you can continue to use the instances that you have already created. |
References
For more information about Elasticsearch Serverless, see What is Elasticsearch Serverless?.
To get started with Elasticsearch Serverless, see Quick Start for Serverless.