Rate Limiting, Quotas, Data Expiry, and Maximums
Rate limiting, Quotas, Data Expiry, and Maximums

Rate Limiting, Quotas, Data Expiry, and Maximums

To ensure quality of service, Haven OnDemand monitors and limits the number of requests, size of requests, and length of retention. These limits fall into several areas as detailed below.

Note: the Haven OnDemand Support team can tweak your plan to accommodate special requirements. If any of these limits restrict your use of the APIs, contact us to discuss premium status.


API Quotas

To prevent users from unexpected charges, there are quotas in a number of areas. You can view the quotas for your account on the Account Management Quotas page. For more information about API resource quotas, see API and Resource Unit Consumption.

If you exceed a quota, the API returns an error message of the following type:

{
  "error": 2010,
  "reason": "A quota has been violated, see additional details",
  "detail": {
    "reason": "Over quota",
    "key": "EXAMPLE123"
  }
}

If your requirements exceed your quotas, please contact us.


Text Index Quotas

The size quotas for your text indexes are determined by the flavor of the text index. Each text index flavor also has a static resource unit cost. For information about the cost and size of each text index flavor, see API and Resource Unit Consumption.

You can view the total units for your account on the Account Management Quotas page.

If your requirements exceed your quotas, please contact us.


Rate Limiting

During normal operation, Haven OnDemand rate limits your requests according to your pricing plan (see Pricing). There are also additional rate limits for the Speech and Video APIs.

If an individual API key or client places a disproportionate amount of load on the system, potentially denying service to other users, a block is placed on that API key for an initial period of six minutes.

In the case of a block, the API returns a 429 HTTP response, with the following error message:

{
    "error": 4030,
    "reason": "Too many requests for this project"
}

If you receive this error, you should stop making requests for 360 seconds to allow the rate limit to expire. After 360 seconds, reduce the rate of requests.

If the API key or client continues to create excessive load, subsequent block periods are increased by a factor of two.

If you have a requirement for a large number of requests, please contact us.

Speech and Video APIs

The following rate limits apply to the Speech Recognition, Detect Scene Changes and License Plate Recognition APIs, as totals for the three together.

  • For free users: two calls per minute and up to 5000 calls per month.
  • For subscribers to any Haven OnDemand paying plan: 30 calls per minute.

If you exceed these rates, the API returns a 429 HTTP response.

If you require higher rate limits, please contact us.

Input Size and Job Duration Limits

To provide a consistent API response experience, there is a maximum size of files that you can pass to the APIs. The default limits for the Early Access accounts are:

File input 1 GB
URL input 100 MB
JSON input 25 MB
Text input 1 MB

Note: HPE Keyview, which is called by a number of APIs including Text Extraction, View Document and Add To Text Index, has a file input limit of 10MB. Be aware of this, as it is well below the limit for inputs that do not require format conversion.

In addition, there can be limits on the length of time taken to run requests.

  • API requests made in sync mode must complete within two minutes.
  • API requests made in async mode must complete within two weeks from the moment of submission.

The Speech Recognition, Detect Scene Changes and License Plate Recognition APIs are also limited to:

  • File length: 30 minutes. The file is truncated beyond this length.
  • Processing time: 2 hours. After this period, the processing terminates and the job returns only what it has completed within that time.

If your requirements exceed these quotas, please contact us.


Data Expiry

Haven OnDemand persists the results of API calls to facilitate asynchronous usage and allow the result of API calls such as storeobject to be used in other APIs. Data stored for these purposes is purged after a minimum of 24 hours.

Haven OnDemand also allows you to see detailed request logs for your account. Request logs are maintained for three months.

If your document has expired and no longer exists, the API returns an error message of the following type:

{
  "error": 5013,
  "reason": "Reference does not exist",
  "reference": "10411270-23ad-4ebb-8f5d-760232298f3b"
}