archive.org /download web serving path

Storage · 1 record · JSON

archive.org /download web serving path (the free CDN) Storage

caveatinstitution?card?Rper-ip-shaping

archive.org/download/<identifier>/<filename> · academic · storage/archive-org-download-web-serving-path · JSON · compare

Free tier	Serving is unmetered and free; no per-file download cap and no egress fee. Practical ceiling is throughput, not quota. unit mixed; metered on unstated
Price	none stated
Per-object cap	not stated
Retention	not stated
Egress	Free / unmetered. Docs: 'high performance downloads are achieved via the archive web infrastructure' — i.e. GET https://archive.org/download/<item>/<file>, which 302s to an iaNNNNNN.us.archive.org datanode. BitTorrent webseeds supplement it.
Identity	institution (placeholder) inferred from the category 'academic'; the source record states no identity cost, so this is a class assumption, not evidence
Limits	per-ip-shaping 1.19 MB/s source per-ip-shaping 1.66 MB/s source Canonical URLs 302 at least once (clients need --location-trusted when sending Authorization); do NOT link iaNNNNNN.us.archive.org directly (nodes move); 429/503 under load; no auth on GET, so everything is world-readable.
Binding	per-ip-shaping one stream is shaped, and a user's throughput is the shaped rate times the number of streams
Interface	rest, rclone, cli rest
Tooling	plain HTTPS GET with Range (measured: 206 + accept-ranges: bytes); ia CLI download w/ --checksum resume; rclone 'internetarchive' reads back through the same front end
Composes with	Parallelise across items — different items can be pinned to different iaNNNNNN datanodes, so aggregate throughput scales with item count rather than with connections to one item. Put Cloudflare in front (replay responses carry cache-control: max-age=1800) so repeated fetches of the hot working set never touch IA at all.
Reliability	Throughput is the binding constraint, not quota: measured 1.19 MB/s on one 90s stream and 1.66 MB/s aggregate across 8 parallel range requests to one datanode from one IP (~145 GB/day) — against a local link calibrated at 27.6 MB/s, so this is server-side shaping, not the client.
Notes	Serving is unmetered and free; no per-file download cap and no egress fee. Practical ceiling is throughput, not quota.
Evidence rung	reported One page read moves it to documented.
Checked	2026-10-21 (inherited); stale on 2026-11-04
Parsed numbers	`$.limits[0].value` limit:per-ip-shaping 1.19 MB `$.limits[1].value` limit:per-ip-shaping 1.66 MB

Evidence (4)

date	source	what it proved
2026-10-21	archive.org/developers/ias3.html	'Fast GET downloads ... high performance downloads are achieved via the archive web infrastructure: curl --location http://archive.org/download/sam-s3-test-08/demo-intro-to-k.pdf'
2026-10-21	archive.org/developers/items.html	(snapshot 2025-11-08) — 'A particular file can always be downloaded from https://archive.org/download/<identifier>/<filename>'
2026-10-21	withheld	probe: curl -r 0-1048575 web.archive.org replay of a 406MB ISO -> HTTP 206, content-range honoured, accept-ranges: bytes, cache-control: max-age=1800
2026-10-21	withheld	probe: 8 parallel 5MB range GETs to one replay target -> 40,000,000 bytes in 23s = ~1.7 MB/s aggregate; single 90s stream -> 1.19 MB/s; Cloudflare 50MB control download -> 27.6 MB/s

Unknown or withheld values (8)

$.evidence[2].url private-url
$.evidence[3].url private-url
$.free_tier.notes process
$.identity_cost.kind identity-defaulted
$.per_object_cap.notes not-extracted
$.per_object_cap.stated not-extracted
$.retention.notes not-extracted
$.retention.stated not-extracted

Built from commit 872967ae41, data as of 2026-10-27. Every value links to its source; freshness badges are computed in your browser. How to read this.