Endpoints and quotas for Amazon OpenSearch Memory
This topic lists the service endpoints, quotas, and limits for Amazon OpenSearch Memory. You can
request an increase to a quota that is marked as adjustable. Unless noted otherwise, a
request that exceeds a limit fails with ValidationException. A
CreateSpace or UpdateSpace request that would exceed an
account quota, such as the number of strategies in a space, fails with
ServiceQuotaExceededException.
Service endpoints
| Region name |
Region |
Control plane endpoint |
Data plane endpoint |
Protocol |
| US East (N. Virginia) |
us-east-1 |
osm.us-east-1.api.aws |
sp-XXXXXXXXXXXX.osm.us-east-1.on.aws |
HTTPS |
| US West (Oregon) |
us-west-2 |
osm.us-west-2.api.aws |
sp-XXXXXXXXXXXX.osm.us-west-2.on.aws |
HTTPS |
| Asia Pacific (Tokyo) |
ap-northeast-1 |
osm.ap-northeast-1.api.aws |
sp-XXXXXXXXXXXX.osm.ap-northeast-1.on.aws |
HTTPS |
| Europe (Frankfurt) |
eu-central-1 |
osm.eu-central-1.api.aws |
sp-XXXXXXXXXXXX.osm.eu-central-1.on.aws |
HTTPS |
Space and tagging operations use the control plane endpoint. Event and memory operations
use the endpoint of the space. The AWS SDKs and the AWS CLI choose the endpoint for you.
OpenSearch Memory doesn't have FIPS endpoints.
Spaces
| Name |
Default |
Adjustable |
| Spaces per account in each AWS Region |
50 |
Yes |
| Memory strategies per space |
10 |
No |
Maximum event retention (eventExpiryDays) |
365 days |
No |
Maximum memory retention (memoryExpiryDays) |
Forever |
No |
Events
| Name |
Default |
Adjustable |
| Messages per conversational event |
100 |
No |
| Text per event, across all messages |
5 MiB (UTF-8) |
No |
| Structured data per event |
1 MiB |
No |
| Maximum JSON nesting depth of structured data |
10 |
No |
| Size of a blob document in Amazon S3 |
25 MB |
No |
| Extraction tokens per minute, per account in each AWS Region |
400,000 |
Yes |
| Event IDs per BatchGetEvent request |
10 |
No |
| Session IDs per SearchEvents request |
10 |
No |
| Session IDs per ListMemories request |
10 |
No |
| Actor IDs per ListMemories request |
10 |
No |
Memories
| Name |
Default |
Adjustable |
| Memory content per CreateMemory request |
16,000 characters |
No |
| Memory content per UpdateMemory request |
16,000 characters |
No |
| Strategy selectors per SearchMemories request |
5 |
No |
| Strategy selectors per ListMemories request |
5 |
No |
| Strategy names per selector |
20 |
No |
| Name |
Default |
Adjustable |
| Distinct metadata keys per space |
50 |
Yes |
Search requests
| Name |
Default |
Adjustable |
Size of query and sort together |
64 KB |
No |
Clauses per query, counting function_score functions |
32 |
No |
Nesting depth of bool queries |
5 |
No |
Values per terms or ids query |
100 |
No |
Decay functions per function_score query |
3 |
No |
Distinct neural query texts per query |
1 |
No |
k in a neural query |
100 |
No |
Event IDs in a terms query on event_ids in a SearchEvents request |
10 |
No |
Results per page (maxResults) |
100 |
No |
| Response size of a SearchMemories request |
5 MiB |
No |
| Response size of a SearchEvents request |
5 MiB |
No |
Request rates
The following table lists the request rate for each operation, per account in each
AWS Region. A request that exceeds the rate fails with ThrottlingException.
The AWS SDKs retry throttled requests automatically. If your application receives
ThrottlingException, retry with exponential backoff. For more information,
see Error retries and exponential backoff.