Ofni Archiver

Transfer Rate

0.00 MB/s 0 Mbps

Network Idle

Worker Threads

0 / 4 Parallel Chunks

Overall Progress

0.0% 0 GB / 0 GB

Estimated Time (ETA)

--:-- Avg: 0 MB/s

Queue Status: Standby

Drag & drop large archive files here

Select or drop one or more files, including files over 20 GB. Data is streamed in S3 multipart chunks.

AWS Lambda Function URL

Multipart uploads go directly from browser to S3

Lambda Function URL

The function URL is saved in this browser. Use HTTPS.

Storage class

S3 Glacier Deep Archive

Chosen on the server when multipart upload is initiated.

Deep Archive retrieval speed

Temporary restored copy (days)

Approximate us-east-1 retrieval charges; temporary Standard storage and download transfer are extra. The original stays in Deep Archive at the same bucket/key; S3 keeps a temporary readable copy there for the selected days.

Connect the Lambda URL and refresh to list archives.

Full at-rest checksum audit

Checks all Deep Archive objects under archives/ without restoring or downloading them. May take about a week; about $4/TB processed plus $0.25 per job.

No checksum audit started.

Upload Queue 0

No files queued. Drop archives above to configure parallel chunks.

Parallel Hardware Tuner

Multi-Thread

Parallel Active Workers: 4 Workers

1 (Sequential) 8 (High Perf) 16 (Max Thread)

S3 Block Chunk Size:

Block Map Status Legend

Pending Block

Uploading (In-Flight)

Completed

Failed / Retry

Technical Architecture & Scaling

Web Workers (Multi-CPU Parallelism): Browser UI threads freeze when processing multi-gigabyte blobs. We instantiate a pool of Dedicated Web Workers that slice archive buffers off-main-thread and execute network requests in true parallel across CPU cores.

Integrity checks: S3 ETags are used to complete and cross-check multipart parts. A multipart ETag is not a whole-file checksum; this version does not download the archive or calculate a full-file hash.

Memory Efficiency: File slices (`File.slice()`) generate lightweight `Blob` references without loading entire 5GB archives into JavaScript heap RAM, preventing out-of-memory browser crashes.