Adjust service quotas for search-enhanced applications-Elasticsearch(ES)-阿里云帮助中心

Prerequisites

You have created a search-enhanced application (Version 8.17).

Go to the application details page

Log on to the Elasticsearch Serverless console. In the top menu bar, switch to the destination region.
In the navigation pane on the left, click Application Management. Click the name of your application to go to its details page.
On this page, you can adjust billing quotas and adjust service quotas and storage capacity as needed.

Adjust billing quotas

Adjust the fixed CU quota

When you create an application, you must configure a fixed CU quota. This quota handles regular workloads and ensures stable processing of service traffic. If the configured fixed CU quota no longer meets your needs, go to the Billing Quota section on the application details page. Click Modify and follow the on-screen instructions to change the fixed CU quota.

Note

You can change the fixed CU quota to specifications such as 2 CU/s, 4 CU/s, 6 CU/s, 8 CU/s, or 16 CU/s. For more information, see CU selection guide.

Adjust the memory usage quota for vector data

The memory usage quota for vector data in an application is the same as the fixed CU quota by default. If this quota does not meet your business needs, go to the Billing Quota section on the application details page. Click Modify and follow the on-screen instructions to change it.

Note

The default memory usage provided by the system is free of charge. Usage that exceeds the default value is billed based on actual usage.
The maximum adjustable memory usage quota varies depending on the fixed CU quota. For more information, see Memory for usage that exceeds the quota. If your business requires more memory than the quota allows, increase the fixed CU quota to obtain more memory.

Adjust service quotas and storage capacity

Go to the application details page. In the navigation pane on the left, click Service Quota. You can then adjust the relevant quotas and index storage capacity as needed. Adjustments to quotas and storage capacity require a request. After you submit a request, you can view the request details and approval status on the Application Record tab.

Adjust service quotas

On the Quota Overview tab, search for a quota item by name or find it by category, such as query or write. Click Modify Quota to adjust the running value of the quota item as needed. After adjusting the quota, click Submit Changes and follow the on-screen instructions to submit the modification request.

Note

Currently, you can only modify quotas by submitting a request. Request resources based on your actual business needs. After you submit a request, go to the Application Record tab to view the request details and approval status.
The approval process takes 1 to 2 business days. For urgent changes, please submit a ticket to contact technical support.
For a list of quota items supported by search-enhanced applications (Version 8.17), see Appendix: List of service quota items.

The following example shows how to adjust the quota for max_storage_per_cu.

Adjust index storage capacity

On the Storage Capacity tab, search for the target index by name. As needed, change the Maximum disk space for index data or Maximum memory for vector data in the index. After making changes, go to the Application Record tab to view the request details and approval status.

Maximum disk space for index data: The maximum disk storage space that a single index can use, in GB. When the index data volume reaches this threshold, the system may trigger alerts, stop write operations, or automatically clear old data. This prevents the index from growing indefinitely, which could exhaust disk space and affect application stability.
Maximum memory for vector data in the index: The maximum memory space that vector data can use, in GB. Exceeding this limit can cause performance degradation or memory overflow. This prevents vector searches from using too much memory and affecting application functions.
According to the calculation method for vector data memory usage, 1,024-dimensional data using int8 quantization supports approximately 1,000,000 documents (that is, the number of documents) per 1 GB of memory. If you require a larger number of documents, you can use bbq quantization, which supports approximately 5,200,000 documents per 1 GB of memory.

Note

After an index is created, the default maximum disk space for data in a single shard is 20 GB. The default maximum memory for vector data in a single shard is 1 GB.
You name your indexes. Index names vary across different applications. The actual names in use prevail.
The approval process takes 1 to 2 business days. For urgent changes, please submit a ticket to contact technical support.

The following example shows how to modify the maximum disk space for the test-alias-index-1746523280 index.

View request and approval details

On the Application Record tab, you can view all request records for quota and storage capacity adjustments. This includes information such as the request type, request time, effective time, approval status, and request content.

In the Operation column, click Revoke to cancel the current change request.
Click Request Item to view the specific content of the current request.
After the request is approved (the status is Completed), the changes take effect automatically.

Appendix: List of service quota items

The following service quota items are supported for search-enhanced applications (Version 8.17).

Note

✅ indicates that the quota value can be changed. 不支持 indicates that the quota value cannot be changed.

Scope	Quota Item	Description	Default Value	Is Value Changeable
Application-level	alias_quota	The maximum number of aliases.	500	✅
Application-level	index_quota	The maximum number of indexes.	500	✅
Application-level	shard_quota	The maximum number of shards.	3000	✅
Application-level	default_app_dense_vector_mem_limit	The maximum total memory usage for vector content across all indexes in a single application, in GB.	5	✅
Shard-level	default_index_dense_vector_mem_limit	The maximum memory usage for vector data in a single shard of a single index, in GB.	1	✅
Shard-level	max_storage_per_shard	The maximum storage usage for a single shard, in GB.	20	✅
Shard-level	max_cu_per_shard	The maximum CU consumption for a single shard in a single query.	2
Index-level	index.mapping.depth.limit	The maximum nesting depth for JSON, in levels.	[1 - 20]	✅
Index-level	index.mapping.field_name_length.limit	The length limit for field names.	[1 - 100]	✅
Index-level	index.mapping.nested_fields.limit	The maximum number of nested fields in a single index.	[1 - 50]	✅
Index-level	index.mapping.nested_objects.limit	The maximum number of nested sub-documents in a single document.	[1 - 100]	✅
Index-level	index.mapping.total_fields.limit	The total number of fields in a single index.	[1 - 2500]	✅
Index-level	index.max_adjacency_matrix_filters	The maximum number of adjacency matrix filters in a single index.	[0 - 100]	✅
Index-level	index.max_docvalue_fields_search	The maximum number of Docvalue_Fields.	[1 - 100]	✅
Index-level	index.max_inner_result_window	The maximum number of results that an internal subquery can return.	[1 - 100]	✅
Index-level	index.max_ngram_diff	The maximum n-gram difference.	[0 - 1]	✅
Index-level	index.max_refresh_listeners	The maximum number of concurrent waits.	[0 - 20]	✅
Index-level	index.max_regex_length	The maximum length of a regular expression.	[0 - 50]	✅
Index-level	index.max_rescore_window	The maximum number of results for rescoring.	[1 - 10000]	✅
Index-level	index.max_result_window	The maximum number of query results.	[1 - 100000]	✅
Index-level	index.max_script_fields	The maximum number of Script Fields.	[1 - 32]	✅
Index-level	index.max_shingle_diff	The maximum shingle difference.	[0 - 3]	✅
Index-level	index.max_terms_count	The maximum number of terms in a single query.	[0 - 1024]	✅
Application-level	max_concurrent_delete_by_query	The maximum number of concurrent Delete_By_Query operations.	3	✅
Application-level	max_concurrent_reindex	The maximum number of concurrent Reindex operations.	3	✅
Application-level	max_concurrent_update_by_query	The maximum number of concurrent Update_By_Query operations.	3	✅
Application-level	max_cu_per_search	The maximum CU consumption for a single query.	20
Application-level	max_doc_size	The size of a single document, in MB.	5	✅
Application-level	max_index_create_and_delete_qps	The QPS for index creation and deletion, in `requests/30s`.	30	✅
Application-level	max_index_metadata_update_qps	The QPS for index metadata updates, in `requests/10s`.	50	✅
Application-level	max_index_template_num	The maximum number of index templates.	50	✅
Application-level	max_pipeline_num	The maximum number of pipelines.	100	✅
Application-level	max_prefix_length	The maximum length of a prefix string allowed in a prefix query.	[0 - 50]	✅
Application-level	max_refresh_qps	The refresh rate, in `requests/s`.	20	✅
Application-level	max_search_throughput	The maximum search throughput, in `MB/s`.	200	✅
Application-level	max_storage_per_cu	The maximum storage ratio per CU. Important Increasing this ratio affects query and write performance. Proceed with caution.	40	✅
Application-level	max_timeout_per_request	The maximum time for a single query, in seconds.	30
Application-level	max_wildcard_length	The maximum length of a wildcard string allowed in a wildcard query.	[0 - 50]	✅
Application-level	max_write_size_per_request	The size of a single write request, in MB.	10	✅
Application-level	search.max_buckets	The maximum number of buckets for a single aggregation.	10000	✅
Application-level	search.max_keep_alive	The maximum keep-alive time for a search, in seconds.	900	✅