Technical Limitations of the Processing Endpoint#
Our processing infrastructure applies the following technical limitations to what and how much may be processed by each individual customer.
Rate-limits#
As a protection against over-saturation of our overall processing capacity we apply rate-limits based on uploaded document pages per single document and certain time-intervals. These limits are applied in order for us to be able to guarantee low processing latency to all our customers at all times.
To prevent running into these rate limits please consider distributing your processing requests evenly over time and avoid uploading large batches of documents at once.
Page based rate-limits apply per tenant and workflow and are independent of the billing-related monthly credits.
Default rate limits for each package#
Limits in pages per...
| Uploaded Document | Second | Minute | Hour | |
|---|---|---|---|---|
| Free | 100 | 10 | 100 | 1000 |
| Paid (maximum value, actual value may vary depending on package) | 250 | 100 | - | - |
If your processing requirements exceed these default rate limits, individual limits can be customized for your tenant as part of the contract details for your purchased processing package.
Detailed Example (click to expand)
Assuming that the free package is used, the following sequence of processing requests could occur (each table row being one uploaded document):
| Upload time | # Pages | Effect | Explanation |
|---|---|---|---|
| 00:01:01 | 1 | Processed | Within rate limits |
| 00:01:02 | 101 | HTTP 429 | Document succeeds limit of 100 pages/document. This document can not be processed and retries will also result in HTTP 429. Since processing is denied, the 101 pages will not be accounted for rate limits not billed quota. |
| 00:01:02 | 5 | Processed | Within rate limits |
| 00:01:02 | 6 | Processed | Within rate limits |
| 00:01:02 | 51 | HTTP 429 (retry in 1s) |
The total number of pages already processed within this second (11 pages at 00:01:02) have surpassed the limit/second (10 pages). Since processing is denied, these 51 pages will not be accounted for rate limits nor billed credits. |
| 00:01:03 | 51 (retry) | Processed | Within rate limits (a new second has started) |
| 00:01:04 | 40 | Processed | Within rate limits |
| 00:01:04 | 1 | HTTP 429 (retry in 1m) |
The total number of pages already processed within this minute (102 pages at 00:01) have surpassed the limit/minute (100 pages). Since processing is denied, the 1 page will not be accounted for rate limits nor billed credits. |
| 00:02:00 | 1 (retry) | Processed | Within rate limits (a new minute has started) |
So in essence, each server-side hour/minute/second has a "page counter" that may exceed the respective limit by an incoming request, however still being accepted for processing. After surpassing the respective rate limit, the server will deny processing any further pages within the same hour/minute/second starting with the next incoming request. Over-use does not transfer to later hours/minutes/seconds but each passing hour/minute/second will completely reset its counter.
Note
The rate limits per time unit are lenient to small processing spikes caused by individual documents with page counts greater than the limit. They are triggered only as soon as the rate limit had been exceeded by the previous request.
Note
The "Free" package by default additionally limits processing capacity to 100 pages per month per workflow for evaluation purposes. If you require higher volumes for individual workflows for proper evaluation, then please get in touch.
File Limits#
Size#
Files uploaded to the processing endpoint are currently limited to 100 MB.
Supported File Types#
- Image formats: JPEG, PNG, GIF, TIFF
- Document formats: PDF
Uploaded images should have a sufficient resolution to support good extraction results. Conversely, lower resolutions lead to faster upload times. We recommend 220 dpi, which makes dimensions of 2600x1800 pixels for an A4 document.