DesignSafe-CI

designsafe-ci.org · Compute, Storage · 3 records · JSON

DesignSafe-CI Data Depot (NSF NHERI / TACC) - no published cap on published datasets Storagecaveatemailno cardRwrite-gated
DesignSafe-CI Computefails: identityinstitutionno cardDwrite-gated
designsafe-ci / nheri data depot (tacc) Storagefails: quotaemail?card?R

DesignSafe-CI Data Depot (NSF NHERI / TACC) - no published cap on published datasets Storage

caveatemailno cardRwrite-gated

www.designsafe-ci.org/ · academic · storage/designsafe-ci-data-depot-no-published-cap-on-published-d · JSON · compare

Free tier	No published cap on published dataset size or file count: 'right now there is no limit to the amount of files that can be published in DesignSafe'. Policies page describes largest published datasets as ~5 TB (CoreTrustSeal submission: 'Publications range up to a maximum of 5TB; seven current published datasets are over 1TB'), while a 2025 Frontiers paper reports individual datasets 'as large as 26' TB. Private working space ('My Data' / 'My Projects') is separate and unlimited in policy terms; uploads of 1 TB+ require a support ticket first. unit TB; metered on unstated
Price	none stated
Per-object cap	not stated
Retention	not stated
Egress	Globus and HTTPS download from the public Data Depot; zip downloads capped at 2 GB per request.
Identity	email email
Limits	write-gated approval source Write-gated twice: a TACC account (free, self-register) and curation 'Publish' approval; datasets must be natural-hazards engineering/social-science research to be in scope.
Binding	write-gated the write is a queue with a person or a rule in it, so the upload rate is set by somebody else
Interface	rest rest
Tooling	Tapis/Agave REST API + dapi Python client (dapi.publications), Globus, JupyterHub, and the Data Depot browser. Public metadata endpoints need no token; the dapi client expects a Tapis token even for public listings, and the unauthenticated /api/files/v2/ media path 404s - so treat the byte-read path as authenticated-or-Globus until proven otherwise.
Composes with	Per published project, and there is no documented per-project ceiling, so the multiplier is projects. NSF-funded continuity: data transfers to a successor awardee and TACC commits to keeping landing pages and DOIs indefinitely.
Reliability	CoreTrustSeal-certified, two-region replication (Corral + Ranch tape). Downloads over ~2 GB need Globus, which is a client install, not a URL.
Notes	No published cap on published dataset size or file count: 'right now there is no limit to the amount of files that can be published in DesignSafe'. Policies page describes largest published datasets as ~5 TB (CoreTrustSeal submission: 'Publications range up to a maximum of 5TB; seven current published datasets are over 1TB'), while a 2025 Frontiers paper reports individual datasets 'as large as 26' TB. Private working space ('My Data' / 'My Projects') is separate and unlimited in policy terms; uploads of 1 TB+ require a support ticket first.
Evidence rung	reported One page read moves it to documented.
Checked	2026-10-21 (inherited); stale on 2026-11-04

Evidence (6)

date	source	what it proved
2026-10-21	designsafe-ci.org/user-guide/curating/faq/	'While right now there is no limit to the amount of files that can be published in DesignSafe...'
2026-10-21	designsafe-ci.org/user-guide/curating/policies/	TACC Corral + Ranch, replication, DOIs, continuity plan
2026-10-21	www.coretrustseal.org/wp-content/uploads/2023/01/20230126-de	'Publications range up to a maximum of 5TB; seven current published datasets are over 1TB'
2026-10-21	designsafe-ci.org/user-guide/managingdata/datatransfer/	'Please submit a ticket to notify us prior to uploading 1TB or more of data'
2026-10-21	www.designsafe-ci.org/api/publications/v2/PRJ-6238/	200, full fileObjs with system 'designsafe.storage.published' and per-file length (e.g. 3019136 bytes)
2026-10-21	www.designsafe-ci.org/api/files/v2/media/system/designsafe.s	... - 404 on every unauthenticated variant tried (media, download, list)

Unknown or withheld values (6)

$.egress.terms partly-held
$.free_tier.notes process
$.per_object_cap.notes not-extracted
$.per_object_cap.stated not-extracted
$.retention.notes not-extracted
$.retention.stated not-extracted

DesignSafe-CI Compute

fails: identityinstitutionno cardDwrite-gated

designsafe-ci.org/user-guide/tools/advanced/hpcallocations/ · GPU cloud · compute/designsafe-ci · JSON · compare

Free tier	A Startup Allocation may request up to 20,000 node-hours annually with fast-track review acted on within two weeks, a Research Allocation up to 100,000 node-hours annually, and an Educational Allocation up to 10,000 node-hours, all provided through NSF funding "at no direct cost to the end user". Unfunded exploratory research is explicitly allowed on a Startup Allocation. Users outside the NSF eligibility categories may buy time from TACC at approximately USD 0.50 per node-hour and storage at approximately USD 60 per TB per year. unit hours; metered on wall-clock; merit-reviewed-allocation
Price	none stated
Per-object cap	not applicable
Retention	not applicable
Egress	not stated
Identity	institution not published
Limits	write-gated The PI is "typically a faculty member or research scientist at a US-based academic institution" and the enumerated alternatives are all US organisations, agencies, school districts, museums or companies; a self-employed person outside the US appears in no category, and foreign collaborators reach the machine only through a US PI's allocation. source The PI is "typically a faculty member or research scientist at a US-based academic institution" and the enumerated alternatives are all US organisations, agencies, school districts, museums or companies; a self-employed person outside the US appears in no category, and foreign collaborators reach the machine only through a US PI's allocation.
Binding	write-gated The PI is "typically a faculty member or research scientist at a US-based academic institution" and the enumerated alternatives are all US organisations, agencies, school districts, museums or companies; a self-employed person outside the US appears in no category, and foreign collaborators reach the machine only through a US PI's allocation.
Interface	unknown inferred from the tooling the provider documents; not probed
Tooling	not stated
Composes with	The purchased route is the composition worth noting: USD 0.50 per node-hour buys a TACC node of 48 to 64 cores, so roughly USD 0.008 to USD 0.01 per core-hour with no affiliation and no review, which is the cheapest way out of the identity gate found and does not depend on the NSF eligibility categories at all.
Reliability	no reliability event recorded
Notes	The account is open to anyone, but the account alone does not reach the HPC queue: portal tools are capped by "built-in constraints that limit the amount of compute time" with no published ceiling, and command-line access needs the allocation. Startup review skips the Allocation Advisory Board and is decided within two weeks, which makes 20,000 node-hours the realistic ask rather than the 100,000. The no-cost extension reported in the July 2025 council message is a horizon fact: the corpus should re-check this record when that extension ends rather than treating the allocation as permanent. The corpus holds no DesignSafe record. Node-hour to core-hour conversion depends on the TACC system assigned, which the page does not name, so no conversion is published here.
Evidence rung	documented
Checked	2026-10-22 (live); stale on 2026-11-21

Evidence (5)

date	source	what it proved
2026-10-22	designsafe-ci.org/user-guide/tools/advanced/hpcallocations/	"A Startup Allocation may request up to 20,000 node-hours annually"; "A Research allocation has a maximum size of 100,000 node-hours annually"; "An Educational allocation may request up to 10,000 node-hours", quoted
2026-10-22	designsafe-ci.org/user-guide/tools/advanced/hpcallocations/	"an additional allocation via DesignSafe is headed by a project PI, typically a faculty member or research scientist at a US-based academic institution"; "foreign collaborators are eligible to make use of that allocation", quoted
2026-10-22	designsafe-ci.org/user-guide/tools/advanced/hpcallocations/	"Allocations are provided through NSF funding at no direct cost to the end user"; purchase rates "approximately $0.50 per node hour" for compute and "approximately $60/TB/year" for storage, dated January 2024 in the text, quoted
2026-10-22	designsafe-ci.org/user-guide/tools/advanced/hpcallocations/	"DesignSafe is broadly available to any researcher or practitioner working in open Natural Hazards Engineering and Social Science research", quoted
2026-10-22	www.designsafe-ci.org/community/news/2025/july/message-from-	services, including HPC resources and data publication, are described as operating during a no-cost extension, quoted

More fields (1)

{
 "wave2": {
  "class": "national-hpc",
  "free_tier_kind": "merit-reviewed-allocation",
  "shape": {
   "vcpu": null,
   "ram_gib": null,
   "gpu_class": null,
   "gpu_count": null
  }
 }
}

Unknown or withheld values (10)

$.egress.metered egress-unknown
$.egress.terms not-stated
$.identity_cost.detail process
$.interface interface-unknown
$.per_object_cap.notes process
$.per_object_cap.stated not-applicable
$.reliability_note none-observed
$.retention.notes process
$.retention.stated not-applicable
$.tooling not-stated

designsafe-ci / nheri data depot (tacc) Storage

fails: quotaemail?card?R

www.designsafe-ci.org/user-guide/curating/policies/ · rejected-provider · storage/designsafe-ci-nheri-data-depot · JSON · compare

Free tier	Publishing is genuinely uncapped — 'Currently we do not impose restrictions on the size of the datasets that can be published ... Largest published datasets in DDR are ~5 TB' — but it is a c unit none; metered on none
Price	none stated
Per-object cap	not recorded
Retention	not recorded
Egress	not recorded
Identity	email (placeholder) not published
Binding	none no constraint of a named kind was recorded; the verdict rests on the quota sentence rather than on a named limit
Interface	unknown
Tooling	not recorded
Composes with	not recorded
Reliability	Publishing is genuinely uncapped — 'Currently we do not impose restrictions on the size of the datasets that can be published ... Largest published datasets in DDR are ~5 TB' — but it is a c
Notes	Publishing is genuinely uncapped — 'Currently we do not impose restrictions on the size of the datasets that can be published ... Largest published datasets in DDR are ~5 TB' — but it is a c
Evidence rung	reported
Checked	2026-10-21 (inherited); stale on 2026-11-04

Evidence (1)

date	source	what it proved
2026-10-21	www.designsafe-ci.org/user-guide/curating/policies/	Publishing is genuinely uncapped — 'Currently we do not impose restrictions on the size of the datasets that can be published ... Largest published datasets in DDR are ~5 TB' — but it is a c

More fields (1)

{
 "dead_end": {
  "failure_class": "quota",
  "why": "Publishing is genuinely uncapped — 'Currently we do not impose restrictions on the size of the datasets that can be published ... Largest published datasets in DDR are ~5 TB' — but it is a c",
  "retry_when": null,
  "first_recorded": "2026-10-21"
 }
}

Unknown or withheld values (15)

$.chaining not-recorded
$.confidence_note withheld
$.egress.terms not-recorded
$.extras.dead_end.retry_when process
$.free_tier.notes process
$.identity_cost.detail defaulted
$.identity_cost.kind identity-defaulted
$.interface interface-unknown
$.interface_detail withheld
$.limits_note process
$.per_object_cap.notes not-recorded
$.per_object_cap.stated not-recorded
$.retention.notes process
$.retention.stated not-recorded
$.tooling not-recorded