Edgecore AS4610: four 1G access switches, 1650 W of PoE that is not 48 x 30 W, and a capacity figure its own datasheet doubles #1555

Merged
jaylouisw merged 1 commit from catalogue/1115-2226 into main 2026-11-15 23:21:34 +00:00
Owner

Adds the Edgecore AS4610 family (bare-metal 1G access switches, ONIE, no NOS) to devices/edgecore/: AS4610-30T, AS4610-30P, AS4610-54T, AS4610-54P - 4 specs.

What the datasheets say

model copper uplinks PoE budget switching capacity max system power
AS4610-30T 24x 10/100/1000BASE-T 4x 1/10G SFP+ - 104 Gbps 76.1 W
AS4610-30P 24x PoE (af/at, 8x UPoE 60 W) 4x 1/10G SFP+ 960 W 104 Gbps 81.72 W
AS4610-54T 48x 10/100/1000BASE-T 4x 1/10G SFP+ - 128 Gbps 92.2 W
AS4610-54P 48x PoE (af/at, 8x UPoE 60 W) 4x 1/10G SFP+ 1650 W 128 Gbps 99.1 W

All four: BCM56340 Helix4, dual-core ARM Cortex-A9 1 GHz, 2 GB DDR3 ECC, 8 MB SPI flash, 4 MB shared buffer, 96 K MACs, 4 K VLANs, 1x 1G RJ-45 management port, serial console, 2x 20G QSFP+ stacking cages, 1RU, two hot-swap AC PSUs.

Discrepancies stated, not smoothed over

  1. Switching capacity. The AS4610-54T R09 and AS4610-54P R08 model datasheets print 256 Gbps full duplex, 190 Mpps. The vendor product pages and every revision of the series specification table print 128 Gbps (and 191 Mpps). 48x1G + 4x10G + 2x20G of full-duplex port bandwidth is 128 Gbps, so 256 Gbps is every port counted in both directions. The catalogue records 128 Gbps and the fabric note on each spec carries the conflict. The 30T/30P 104 Gbps figure agrees across both documents; the model datasheets print 95 Mpps where the series table prints 156 Mpps, so only the capacity is recorded.
  2. 1650 W is not 48 x 30 W. The -54P budget of 1650 W is less than 8 x 60 W + 40 x 30 W (1680 W), and the datasheet is explicit that the AC PSUs are redundant for the system and the network but not fully redundant for PoE or the power budget. The poe note says so; every copper port is modelled at 30 W because the datasheet never enumerates which eight ports are the 60 W UPoE ones.
  3. The 2x 20G QSFP+ stacking cages are not modelled as ports. 20G is not a line rate the importer accepts (SPEEDS has no 20G), and calling them 40G QSFP+ or 10G would state a rate no datasheet gives. They are described in notes, the declared port list is 24/48 + 4 SFP+ + mgmt + console.
  4. All four are end of life since 30 June 2025 (EOL notice, linked from each product page). Specs stay - the hardware is still in the field.

Sources (per spec: sourceUrl = product page; PDFs named in notes, poe.sourceUrl, fabric.note)

Validation

  • node scripts/import-devices.js --apply (isolated PF_INCOMING): 4 specs, 0 rejected, 0 duplicates, 0 capability losses. Two warnings per PoE model (802.3af / 802.3at are descriptive tokens the engine does not implement - kept as declared, same as every other PoE switch in the catalogue).
  • node scripts/audit-port-types.js - exit 0, "No port contradicts its declared type."
  • node scripts/audit-devices.js - "No hard defects."
  • node scripts/syntax-check.js - 325 files, all OK.
  • node --test test/importer-roundtrip.test.js test/catalogue-hygiene.test.js test/import-devices-capabilities.test.js test/faceplate-catalogue.test.js test/doc-catalog-counts.test.js test/poe-agreement.test.js - all pass.
  • node scripts/regen-registry-doc.js - switch count 298 -> 302.

Unverifiable / left out: which eight numbered ports carry UPoE (datasheet shows a block, not numbers); forwarding rate where the two documents disagree (95 vs 156 Mpps, 190 vs 191 Mpps); console flavour varies by airflow variant (RJ-45 or micro-USB), so both are described rather than one declared.

Adds the Edgecore AS4610 family (bare-metal 1G access switches, ONIE, no NOS) to `devices/edgecore/`: **AS4610-30T, AS4610-30P, AS4610-54T, AS4610-54P** - 4 specs. **What the datasheets say** | model | copper | uplinks | PoE budget | switching capacity | max system power | |---|---|---|---|---|---| | AS4610-30T | 24x 10/100/1000BASE-T | 4x 1/10G SFP+ | - | 104 Gbps | 76.1 W | | AS4610-30P | 24x PoE (af/at, 8x UPoE 60 W) | 4x 1/10G SFP+ | 960 W | 104 Gbps | 81.72 W | | AS4610-54T | 48x 10/100/1000BASE-T | 4x 1/10G SFP+ | - | 128 Gbps | 92.2 W | | AS4610-54P | 48x PoE (af/at, 8x UPoE 60 W) | 4x 1/10G SFP+ | 1650 W | 128 Gbps | 99.1 W | All four: BCM56340 Helix4, dual-core ARM Cortex-A9 1 GHz, 2 GB DDR3 ECC, 8 MB SPI flash, 4 MB shared buffer, 96 K MACs, 4 K VLANs, 1x 1G RJ-45 management port, serial console, 2x 20G QSFP+ stacking cages, 1RU, two hot-swap AC PSUs. **Discrepancies stated, not smoothed over** 1. **Switching capacity.** The AS4610-54T R09 and AS4610-54P R08 model datasheets print *256 Gbps full duplex, 190 Mpps*. The vendor product pages and every revision of the series specification table print *128 Gbps* (and 191 Mpps). 48x1G + 4x10G + 2x20G of full-duplex port bandwidth is 128 Gbps, so 256 Gbps is every port counted in both directions. The catalogue records **128 Gbps** and the `fabric` note on each spec carries the conflict. The 30T/30P 104 Gbps figure agrees across both documents; the model datasheets print 95 Mpps where the series table prints 156 Mpps, so only the capacity is recorded. 2. **1650 W is not 48 x 30 W.** The -54P budget of 1650 W is less than 8 x 60 W + 40 x 30 W (1680 W), and the datasheet is explicit that the AC PSUs are redundant for the system and the network but *not* fully redundant for PoE or the power budget. The `poe` note says so; every copper port is modelled at 30 W because the datasheet never enumerates which eight ports are the 60 W UPoE ones. 3. **The 2x 20G QSFP+ stacking cages are not modelled as ports.** 20G is not a line rate the importer accepts (`SPEEDS` has no 20G), and calling them 40G QSFP+ or 10G would state a rate no datasheet gives. They are described in `notes`, the declared port list is 24/48 + 4 SFP+ + mgmt + console. 4. All four are **end of life since 30 June 2025** ([EOL notice](https://www.edge-core.com/eol-notice/2025-june-eol-notice-as4610-series/), linked from each product page). Specs stay - the hardware is still in the field. **Sources** (per spec: `sourceUrl` = product page; PDFs named in `notes`, `poe.sourceUrl`, `fabric.note`) - https://www.edge-core.com/product/as4610-30t/ , /as4610-30p/ , /as4610-54t/ , /as4610-54p/ - AS4610 series datasheet (R09): https://www.edge-core.com/wp-content/uploads/2023/08/AS4610-54T-R09.pdf - AS4610-54P R08: https://www.edge-core.com/wp-content/uploads/2023/08/AS4610-54P-R08.pdf - AS4610-54T R09: https://www.edge-core.com/wp-content/uploads/2023/08/AS4610-54T-R09.pdf - Budget/port table cross-checked against AS4610 series DS R03/R04/R05 mirrors (EPS Global, IBM ICP hardware guide) - all three agree on 960 W / 1650 W and 8x UPoE. **Validation** - `node scripts/import-devices.js --apply` (isolated `PF_INCOMING`): 4 specs, 0 rejected, 0 duplicates, 0 capability losses. Two warnings per PoE model (`802.3af` / `802.3at` are descriptive tokens the engine does not implement - kept as declared, same as every other PoE switch in the catalogue). - `node scripts/audit-port-types.js` - exit 0, "No port contradicts its declared type." - `node scripts/audit-devices.js` - "No hard defects." - `node scripts/syntax-check.js` - 325 files, all OK. - `node --test test/importer-roundtrip.test.js test/catalogue-hygiene.test.js test/import-devices-capabilities.test.js test/faceplate-catalogue.test.js test/doc-catalog-counts.test.js test/poe-agreement.test.js` - all pass. - `node scripts/regen-registry-doc.js` - switch count 298 -> 302. **Unverifiable / left out:** which eight numbered ports carry UPoE (datasheet shows a block, not numbers); forwarding rate where the two documents disagree (95 vs 156 Mpps, 190 vs 191 Mpps); console flavour varies by airflow variant (RJ-45 or micro-USB), so both are described rather than one declared.
Edgecore AS4610: four 1G access switches, 1650 W of PoE that is not 48 x 30 W, and a capacity figure its own datasheet doubles
Some checks failed
CI / Classify the change (pull_request) Successful in 17s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 21s
CI / Test (Node 18) (pull_request) Failing after 3m43s
CI / Test (Node 20) (pull_request) Failing after 3m19s
CI / Server smoke test (pull_request) Has been skipped
CI / Docker build (pull_request) Has been skipped
CI / Test (Node 22) (pull_request) Failing after 3m16s
6ad40daf79
Adds the AS4610-30T/30P/54T/54P bare-metal 1G access switches: 24 or 48
10/100/1000BASE-T ports, 4x 1/10G SFP+ uplinks, 1x 1G RJ-45 management
port, a serial console, and (on the P models) PoE on every copper port
with 8 ports of UPoE at 60 W on a 960 W or 1650 W budget.

The 2x 20G QSFP+ stacking cages are described in the spec notes but not
declared as interfaces: 20G is not a line rate the importer accepts, and
calling them 40G QSFP+ or 10G would state a rate no datasheet gives.

Switching capacity is recorded as 104 Gbps (30T/30P) and 128 Gbps
(54T/54P) - the full-duplex aggregate of the port rates. The AS4610-54T
R09 and AS4610-54P R08 model datasheets print 256 Gbps instead; 256 is
every port counted in both directions, the product pages and the series
table both say 128, so 128 is what the catalogue carries.

All four are end-of-life since 30 June 2025; the specs stay because the
hardware is still in the field.
jaylouisw force-pushed catalogue/1115-2226 from 6ad40daf79
Some checks failed
CI / Classify the change (pull_request) Successful in 17s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 21s
CI / Test (Node 18) (pull_request) Failing after 3m43s
CI / Test (Node 20) (pull_request) Failing after 3m19s
CI / Server smoke test (pull_request) Has been skipped
CI / Docker build (pull_request) Has been skipped
CI / Test (Node 22) (pull_request) Failing after 3m16s
to 8023164d9d
Some checks failed
CI / Classify the change (pull_request) Successful in 16s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 21s
CI / Test (Node 20) (pull_request) Has started running
CI / Server smoke test (pull_request) Has been skipped
CI / Docker build (pull_request) Has been skipped
CI / Test (Node 18) (pull_request) Failing after 3m33s
CI / Test (Node 22) (pull_request) Failing after 10m41s
2026-11-15 22:57:42 +00:00
Compare
Author
Owner

Rebased onto main (d013e7c2) — the only conflict was docs/DEVICE_REGISTRY.md, which is generated. I took main's copy and regenerated with node scripts/regen-registry-doc.js rather than hand-merging: 2,074 -> 2,078 models, switch 306 -> 310 (the four AS4610s). Nothing hand-edited. Verified: doc-catalog-counts 24, catalogue-hygiene 25, faceplate-catalogue 14, device-aliases 13, missing-devices 66, apply-research 10, import-devices-capabilities 29 — all passing, 0 failed. Head is now 8023164d, mergeable. No CI run has been dispatched for it yet.

Rebased onto main (d013e7c2) — the only conflict was docs/DEVICE_REGISTRY.md, which is generated. I took main's copy and regenerated with `node scripts/regen-registry-doc.js` rather than hand-merging: 2,074 -> 2,078 models, switch 306 -> 310 (the four AS4610s). Nothing hand-edited. Verified: doc-catalog-counts 24, catalogue-hygiene 25, faceplate-catalogue 14, device-aliases 13, missing-devices 66, apply-research 10, import-devices-capabilities 29 — all passing, 0 failed. Head is now 8023164d, mergeable. No CI run has been dispatched for it yet.
Author
Owner

This branch conflicts with main, so it was not rebased:

devices/edgecore/as4610-30p.json devices/edgecore/as4610-30t.json devices/edgecore/as4610-54p.json devices/edgecore/as4610-54t.json docs/DEVICE_REGISTRY.md research/incoming/devices-edgecore-as4610.json 

A conflict is a decision about behaviour rather than a mechanical fix, so this is left for a human. Rebasing by hand and force-pushing will let the automation pick it up again.

This branch conflicts with `main`, so it was not rebased: ``` devices/edgecore/as4610-30p.json devices/edgecore/as4610-30t.json devices/edgecore/as4610-54p.json devices/edgecore/as4610-54t.json docs/DEVICE_REGISTRY.md research/incoming/devices-edgecore-as4610.json ``` A conflict is a decision about behaviour rather than a mechanical fix, so this is left for a human. Rebasing by hand and force-pushing will let the automation pick it up again.
jaylouisw scheduled this pull request to auto merge when all checks succeed 2026-11-15 23:21:18 +00:00
jaylouisw force-pushed catalogue/1115-2226 from 8023164d9d
Some checks failed
CI / Classify the change (pull_request) Successful in 16s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 21s
CI / Test (Node 20) (pull_request) Has started running
CI / Server smoke test (pull_request) Has been skipped
CI / Docker build (pull_request) Has been skipped
CI / Test (Node 18) (pull_request) Failing after 3m33s
CI / Test (Node 22) (pull_request) Failing after 10m41s
to dd4afed659
Some checks failed
PR rebase / Rebase onto main and arm auto-merge (pull_request) Failing after 10m12s
CI / Classify the change (pull_request) Failing after 10m19s
CI / Test (Node 18) (pull_request) Has been cancelled
CI / Test (Node 20) (pull_request) Has been cancelled
CI / Test (Node 22) (pull_request) Has been cancelled
CI / Server smoke test (pull_request) Has been cancelled
CI / Docker build (pull_request) Has been cancelled
2026-11-15 23:21:23 +00:00
Compare
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge!1555
No description provided.