Limits on the system
Item | Limit |
Number of instances for each user | Not limited. |
Total number of documents for each user | Theoretically not limited. The value is calculated based on the storage capacity quota. |
Total number of page views (PVs) for each user | Theoretically not limited. The value is calculated based on the maximum number of logical computing units (LCUs) that are available for the user. |
Character encoding system | UTF-8 is supported. |
Limits on instances
Item | Limit | |
Cluster type | Shared cluster | Exclusive cluster |
Length of an instance name | 30 characters | |
Length of a field name | 30 characters | |
Length of a sort expression name | 30 characters | |
Number of secondary tables | 10 | |
Number of fields | 256 | 512 |
Length of a source table name | 16 characters | |
Length of an index field name | 64 characters | |
Levels of secondary tables that can be joined to a primary table | Two levels | Three levels |
Number of fields of the INT type | 256 | |
Number of fields of the TIMESTAMP type | 4 | |
Number of fields of the GEO_POINT type | 2 | |
Number of fields of the LITERAL type. Fields of the LITERAL type cannot be used to create composite indexes. | 256 | |
Number of fields of the TEXT or SHORT_TEXT type | 32 | 64 |
Number of indexes for numerical analysis | 4 | 8 |
Number of composite indexes | 4 | 8 |
Number of fields in a composite index | 8 | |
Number of indexes that are created on fields of the INT type for numerical analysis | 4 | 8 |
Number of indexes that are created on fields of the GEO_POINT type for numerical analysis | 2 | |
Number of indexes that are created on fields of the TIMESTAMP type for numerical analysis | 4 | |
Number of indexes that are created on a single field of the TEXT or SHORT_TEXT type | 4 | 8 |
Number of indexes that are created on a single field of the LITERAL type | 256 | |
Number of indexes that are created on fields of the INT type for keyword analysis | 256 | |
Number of vector indexes | 0 | 2 |
Total number of data sources | 20 | |
Limits on fields
Item | Limit |
INT64 | -2^63~2^63-1 |
FLOAT | +/-3.40282e+038 |
DOUBLE | +/-1.79769e+308 |
LITERAL | A field of this type can be up to 65,535 bytes in length. |
TEXT | A field of this type can be up to 65,536 words in length. |
SHORT_TEXT | A field of this type can be up to 100 bytes in length. If a field exceeds 100 bytes in length, the field is truncated. |
LITERAL_ARRAY | A field of this type can be up to 65,535 bytes in length. Each built-in delimiter between field elements is 2 bytes in length. If a field exceeds 65,535 bytes in length, the field is truncated to the last complete element to fit within the maximum length. A large number of elements in a field of this type require high query performance of the system. We recommend that you construct each field with no more than 100 elements. |
INT_ARRAY, FLOAT_ARRAY, and DOUBLE_ARRAY | If fields of these types are configured as attribute fields, each field can contain up to 65,535 elements. We recommend that you construct each field with no more than 100 elements. |
Limits on sort expressions
Item | Limit |
Number of rough sort expressions | 30 |
Number of fine sort expressions | 30 |
Number of feature function items that can be set for each rough sort expression | 4 |
Limits on the search result summary
Item | Description | Limit |
Length of a segment | The maximum length of a summary. | 1 to 300 bytes |
Number of segments | The number of segments that are required in a summary. | 1 to 5 |
Limits on application-level data push
Item | Limit |
Maximum number of documents that you call the API to push each time | 1,000. To provide better performance, we recommend that you push 100 documents each time. We also recommend that you package the documents to be pushed. |
Maximum number of times that you call the API to push data per second | 500. If you have pushed data 500 times in a second, a new request fails, and the error message "rate exceeds app quota" is displayed. |
Maximum total data size of each API request | 2 MB. If the total data size of an API request exceeds 2 MB, the request fails, and the error message "rate exceeds app quota" is displayed. |
Maximum total data size of API requests per second | 2 MB. If the total data size of API requests in a second exceeds 2 MB, a new request fails, and the error message "rate exceeds app quota" is displayed. |
Maximum synchronization rate for incremental data in ApsaraDB RDS or PolarDB data sources | 2 MB/s. If the synchronization rate exceeds 2 MB/s, the system automatically limits the rate and delays the incremental data synchronization. |
Maximum number of updates (TPS) on primary and secondary tables that are triggered by database updates or API operation calls | 1,500. If the number of updates exceeds 1,500, the incremental data synchronization is delayed. |
Maximum size of a document | 1 MB. If the size of a document exceeds 1 MB, the document fails to be synchronized, and the error message "The total size of documents exceed" is displayed. |
Latency in incremental data synchronization | After a number of documents in the primary table are pushed to OpenSearch, 90% of the documents can be searched in 10 seconds, and 99% of the documents can be searched in 10 minutes. For more information about data synchronization of secondary tables, see Data synchronization latency caused by multi-table joins. |
● Exceeding the upper limit of data push delays the incremental data synchronization.
● A large number of secondary table updates delay the primary table synchronization. The system does not guarantee the number of updates on secondary tables per second, and limits the update rate of secondary tables in real time based on the latency.
● If the used storage capacity exceeds the application quota, the system automatically discards the request for API data push and requires data push again after a scale-out.
● If the used storage capacity exceeds the application quota, the system automatically stops synchronizing incremental data from ApsaraDB RDS or PolarDB data sources. After a scale-out, the system automatically resumes incremental data synchronization from the point of time when it stops synchronization.
● If the amount of incremental data generated is more than the amount that can be synchronized in real time, the synchronization is delayed, and some incremental data may not be synchronized to the OpenSearch application. In this case, the reindexing cannot be complete.
● The preceding usage notes apply to both the Standard Edition and Advanced Edition.
● The upper limit of data push can be properly adjusted for applications in exclusive clusters.
● If an ApsaraDB RDS or PolarDB data source is configured and the data synchronization mode is automatic synchronization, OpenSearch ensures the stability of the synchronization but does not guarantee low synchronization latency. If your business requires low synchronization latency, we recommend that you use a change tracking instance of Data Transmission Service (DTS) to synchronize data. For more information, see Configure DTS real-time synchronization.
Non-printable reserved characters that cannot be contained in data to be pushed
Code | Display format in Emacs or Vi |
"\x1E\n" | ^^ |
"\x1F\n" | ^_ |
"\x1C" | ^\ |
"\x1D" | ^] |
"\x03" | ^C |
If a value of a field includes the preceding non-printable characters, the record is not synchronized to an application.
Non-printable characters such as \u0002 and \u0003 are automatically converted to \t.
The engine supports the following character sets:
CJK_1 (\xe2[\xba-\xbf][\x80-\xbf])
CJK_2 ([\xe3-\xe9][\x80-\xbf][\x80-\xbf])
FULLWIDTH_1 (\xef[\xbc-\xbe][\x80-\xbf])
FULLWIDTH_2 (\xef\xbf[\x80-\xaf])
Limits on searches
Item | Limit |
Maximum length of each clause, except filter clauses | 1 KB before encoding. If the length of a clause exceeds 1 KB, an error message is displayed and the request is discarded. |
Maximum length of each filter clause | 4 KB before encoding. If the length of a clause exceeds 4 KB, an error message is displayed and the request is discarded. |
Maximum number of results returned on one page for a single request | 500 |
Maximum number of results returned for a single search, including multiple requests for paged queries | 5,000 |
Maximum number of documents involved in rough sort | 1,000,000 |
Maximum number of documents involved in fine sort | Default value: 200 |
● For traditional searches, the purpose is to retrieve the most matched results in the shortest period of time. Therefore, the number of documents that can be contained in results is limited. For example, the results of a search can contain up to 5,000 documents. However, in some scenarios, you may need more results for analysis. In this case, you can perform scroll queries to obtain more search results.
● If search requests have consumed all LCUs that are available for the application, new requests are discarded and the error message "code:6015,message:Dropped by over compute resource quota" is displayed.
Limits on reindexing
Item | Limit |
Maximum synchronization rate | 2MB/s |
Note: For more information, see How is data synchronized in OpenSearch? | |
● The upper limit can be properly adjusted for applications in exclusive clusters.
● By default, the reindexing automatically fails if it cannot be complete within three days.
● If the reindexing fails after a long period of time or the latency in data synchronization is high, submit a ticket for technical support.