Question
Tokio Marine HCC
GB
Last activity: 24 Aug 2025 21:32 EDT
Clarification on File Upload, Download, and Delete Behavior with External S3 Integration
Hello Pega Support Team,
We are currently testing large file handling with a Constellation-based application (Pega version 8.7.6), which is integrated with an external Amazon S3 bucket. I would like your guidance on the following observations and questions:
-
Upload Behavior
-
We have set the DSS value to
1GB. -
Uploads of large files (e.g., ~300 MB) are working successfully without any issues.
-
Can you please confirm the mechanism used when uploading to S3? Specifically, is the file stream sent through the Pega application server and then uploaded to S3, or is it handled directly between the client browser and the S3 bucket (for example via pre-signed URLs)?
-
-
Download Behavior
-
While uploads succeed, downloads of similarly sized files are timing out.
-
Could you explain how the download is managed? Is the file first pulled from S3 into the Pega app server and then streamed to the client browser, or does Pega support direct browser-to-S3 download links?
-
What configurations or product capabilities are available to ensure that the maximum supported download size is aligned with the upload size (up to the DSS limit of 1GB)?
-
-
Delete Behavior
Hello Pega Support Team,
We are currently testing large file handling with a Constellation-based application (Pega version 8.7.6), which is integrated with an external Amazon S3 bucket. I would like your guidance on the following observations and questions:
-
Upload Behavior
-
We have set the DSS value to
1GB. -
Uploads of large files (e.g., ~300 MB) are working successfully without any issues.
-
Can you please confirm the mechanism used when uploading to S3? Specifically, is the file stream sent through the Pega application server and then uploaded to S3, or is it handled directly between the client browser and the S3 bucket (for example via pre-signed URLs)?
-
-
Download Behavior
-
While uploads succeed, downloads of similarly sized files are timing out.
-
Could you explain how the download is managed? Is the file first pulled from S3 into the Pega app server and then streamed to the client browser, or does Pega support direct browser-to-S3 download links?
-
What configurations or product capabilities are available to ensure that the maximum supported download size is aligned with the upload size (up to the DSS limit of 1GB)?
-
-
Delete Behavior
-
File deletions are working correctly.
-
Could you confirm how the delete operation is executed? Is the delete request processed directly against the S3 bucket, or is it routed through the Pega application server first?
-
-
Performance Impact
-
If downloads and deletes are routed via the Pega app server, what are the performance implications for the application when handling large file transfers and operations?
-
Are there recommended configurations (timeouts, streaming settings, or alternative approaches) to reduce impact on application performance?
-
Your clarification will help us finalize our design for handling large files and ensure we are aligned with Pega best practices.
HCA Healthcare
US
In 8.7.6 with an S3 repository, uploads go from the browser to the Pega app node (multi-part form), land in the temp folder, then Pega writes them to the configured repository (DB or S3) — there’s no built-in direct browser-to-S3 upload unless you custom-implement presigned URLs. Your 1 GB DSS is fine; note that for 8.5+ the default cap is ~1.5 GB and the hard limit is 2 GB, so anything above that will be rejected regardless of the DSS. Downloads are proxied: Pega reads the file from S3 and streams it to the browser using sendFile; out-of-the-box you don’t get a raw S3 “direct” link without going through Pega authentication. Large downloads timing out usually means a load balancer/app-server/browser timeout is expiring while Pega is proxying the stream; ensure streaming via sendFile (pyStream) and raise idle/response timeouts across LB, app server connectors, and any proxies. If you want true large-file performance, generate short-lived S3 presigned GET/PUT URLs in Pega and let the browser talk to S3 directly; this avoids tying up Pega threads and usually fixes timeout issues for big files.Deletes work as you’d expect: Pega validates privileges and then calls the repository delete API (D_pxDelete), which executes the delete in S3. Performance-wise, proxying big transfers through app nodes consumes requestor threads and network bandwidth; if you must proxy, scale web nodes and keep generous timeouts/keep-alives. For configuration hygiene: keep your DSS upload limit consistent with infra limits, use streaming (not full buffering), bump LB/proxy/app-server timeouts to match 1 GB transfers, and prefer the presigned-URL pattern when you expect frequent 100–1000 MB files.