Transfer Capacity¶

Lenzeye is built for volume — thousands of files, hundreds of gigabytes, in a single session.

Validated Capacity (Production)¶

Limit	Value	Notes
Single session files	2,630+ validated	No upper limit enforced
Single session duration	6h 56min	No timeout
Single session data	250 GB+	Limited only by storage plan
Single file size	No hard limit	Tested beyond 50 GB
Part size	10 MB	Fixed chunk size for multipart
Concurrent uploads (plain)	Browser-dependent	No server-side limit for presigned path
Concurrent encrypted uploads	4 simultaneous	Bounded by `BoundedSemaphore(4)`

Lenzeye's architecture is designed so that capacity is limited by storage plan, not by the server:

Plain uploads go browser → S3 directly. Server only handles metadata. No server-side bottleneck.
Encrypted uploads pass through the server for encryption. The BoundedSemaphore(4) caps concurrent encryption threads at 4 to bound memory, but queues additional requests rather than rejecting them.
S3 Multipart supports up to 10,000 parts per object. At 10 MB each, that's 100 GB per file maximum.

Plan	Storage Limit
Free	25 GB
Paid	Higher (plan-dependent)

Storage is checked before upload begins — if your storage is full, the upload is blocked before any data is transferred.

The Gunicorn configuration (1 worker, 6 threads) supports 6 concurrent HTTP connections. For upload-heavy workloads:

Each plain upload thread is lightweight (just generating presigned URLs)
Each encrypted upload thread holds 10 MB in memory for the duration of encryption
The semaphore ensures at most 4 are encrypting simultaneously, others wait in queue

For higher concurrency (10+ simultaneous heavy users), scaling to a higher Render plan with more RAM is the path.

With S3 Multipart's 10,000-part limit and 10 MB chunks: