US 7995580

I'll search for the specific patent number and any associated litigation.

Let me dig into the claim set and any litigation records specifically.

Let me try to retrieve the actual claim text for this specific patent.

Bottom line

US 7,995,580 B2 is a real, now‑expired Foundry Networks patent in the big "high‑performance network switch" family (the same family as US 7,236,490, US 7,596,139, US 8,514,716, US 8,619,781, US 8,964,754, US 9,030,937). I found no CAFC 2026 docket entry, appeal, or 2026 litigation involving 7,995,580 in any of the searches I ran. I also could not retrieve verified verbatim claim text through the tools available, so I flag the claim discussion below accordingly.

1. Bibliographic data

Uncertainty notes: the claim count, the exact claim text, whether the '580 carries a terminal disclaimer (siblings in the family do), and the identity of the prosecuting firm for the 2009 application are all things I could not verify with the sources reached. Treat them as unconfirmed.

2. Classifications

H04L47/6225 (fixed service order, e.g., round robin), H04L45/74, H04L49/153 and H04L49/1538 (ATM switching fabrics having parallel switch planes; cell slicing), H04L49/25 (routing in a switch fabric), H04L49/3063 (pipelined peripheral units), H04L49/901 (buffering using storage descriptors/pointers), H04L49/352 (Gigabit Ethernet switching).

3. Abstract (as reported for the '580)

"A backplane interface adapter with error control and redundant fabric for a high-performance network switch. The error control may be provided by an administrative module that includes a level monitor, a stripe synchronization error detector, a flow controller, and a control character presence tracker. The redundant fabric transceiver of the backplane interface adapter improves the adapter's ability to properly and consistently receive narrow input cells carrying packets of data and output wide striped cells to a switching fabric."

(Note: this is the abstract text carried by the '580 family. The earlier 2002 publication of the parent, US 2002/0105966 A1, had only the final sentence — the error‑control sentence was added in the later continuation filings.)

4. What the disclosure covers

The specification (which I do have in full) describes a switch in which blades are coupled to a switching fabric by serial pipes; data is carried in wide striped cells spread across five stripes (5 × 10 Gbps = 50 Gbps per slot), with in‑band control information in each stripe (K0 = start of cell, K1 = end of packet, K2 = used in the re‑synchronization sequence). A backplane interface adapter (BIA) has three traffic paths: local→fabric, fabric→local, and a local cut‑through path. The error‑control content centers on:

• an administrative module with a level monitor, stripe synchronization error detector, flow controller, and control character presence tracker (FIG. 28A);
• monitoring FIFO levels in stripe receive synchronization queues to detect overflow/underflow on a given stripe from a given source (FIG. 31), initiating a re‑synchronization routine;
• a K2 synchronization sequence sent across all lanes for a predetermined number of cycles (FIG. 34B, routine 29/30);
• redundant fabric transceivers (RFTs) on ingress/egress blades, coupling to two switching blades, with a multiplexer, downlink/uplink transceivers, and a register module — plus synchronous (time‑domain‑multiplexed) and asynchronous flow‑control implementations (FIGS. 33A–37).

5. Independent claims — plain language

Important caveat: the authoritative patent text supplied to me is truncated before the claims section, and my searches did not return the '580 claim set. I therefore cannot give you the verbatim independent claims, and I will not reconstruct them, because doing so would risk fabricating claim language.

What I can say with confidence at the disclosure level is that the '580 (as the error‑control/redundant‑fabric continuation) is directed to subject matter that the specification expressly frames as inventive, namely:

1. Stripe synchronization error detection — sorting data received at a receiving slot by stripe and source into data structures, monitoring the levels of those structures, and detecting overflow/underflow on a respective stripe from a particular source (where "source" may be a slot or a source packet processor), then triggering a re‑synchronization routine.
2. Re‑synchronization of striped traffic — sending a common character (K2) across all lanes for a predetermined number of cycles, evaluating the control characters arriving at the stripe receive synchronization queues, and detecting an in‑synch condition indicating the queues have been cleared.
3. Out‑of‑synchronization flow management — monitoring stripe‑receive‑synchronization queue levels, determining whether an out‑of‑syn condition exists, and initiating the re‑synchronization routine.
4. A redundant switching system / redundant fabric transceiver — two switching blades each with cross points for respective stripes, ingress/egress blades coupled to both via the backplane, and RFTs (with multiplexer, downlink and uplink transceivers, and an optional register module holding configuration/parameter information) that can switch traffic between the cross points of the two switching blades to provide redundancy.

If you need the literal claim language (which is what matters for infringement/validity), pull it from the granted patent PDF at https://patents.google.com/patent/US7995580/en ("Claims" section) or USPTO Patent Center for application 12/400,594 — I was not able to reach that section with the tooling available here.

6. CAFC 2026 dockets

• I found no Federal Circuit case in 2026 (or any year) captioned to US 7,995,580, and no 2026 docket entry listing it. Searches for the number against CAFC 2026 materials returned unrelated decisions (e.g., Enviro Tech v. Safe Foods, No. 2024‑2160; ironSource v. Digital Turbine, No. 2025‑1218) and nothing touching the '580.
• This is consistent with the record: the '580 is marked Expired – Fee Related with an anticipated expiration of May 15, 2021, so an active 2026 appeal is unlikely (a suit filed in 2026 could at most reach back‑damages conduct, and no such suit appeared).
• Caveat: I searched public sources (Google Patents, Justia, USPTO.report, FreePatentsOnline, Unified Patents, and general web results containing CAFC 2026 summaries). I did not have direct, live access to a USPTO PAIR/Patent Center docket or a CAFC docket system, so I cannot certify that no filing exists — only that none appeared.

7. Closing uncertainty statement

Verified from the authoritative full text and consistent public records: title, application number, filing/grant dates, priority chain, inventor list, assignee history, and the technical content. Not verified: the '580's claim count and exact claim wording, the terminal‑disclaimer status of this particular patent, and the absence of any docket activity in fee‑based docket systems.

I'll search for litigation specifically involving US 7,995,580.

The first searches returned matches for other patents (Woodway's 9,039,580, etc.), not US 7,995,580. Let me search more precisely.

Bottom line: I found no litigation specifically involving US 7,995,580

I ran targeted searches for the literal identifier 7995580 / 7,995,580 against patent‑litigation sources (Justia dockets, CourtListener/RECAP, PTAB/PTABCTS filings, Docket Alarm, unifiedpatents.com, RPX/insight, general web). No case — district court, ITC, or PTAB — appeared in which US 7,995,580 is named as an asserted/patented patent. Every "580" hit returned was a different patent, so under the strict rule I am not attributing any of them to this patent.

Note on dates: the task header says April 26, 2026, while the system date supplied to me is 2026‑09‑28. I flag the discrepancy; it does not change the result (no 7,995,580 docket appeared under either date).

1. What the searches actually returned (and why none of it is '580)

I also note the fetch of https://patents.google.com/patent/US7995580/en supplied in the record contains no litigation section and no "Litigation" entries — consistent with the absence of any found case.

2. Family‑level enforcement context (flagged as unverified for '580)

The '580 sits in the Foundry Networks / Brocade "high‑performance network switch" family (US 7,236,490; 7,596,139; 8,514,716; 8,619,781; 8,964,754; 9,030,937), and Foundry/Brocade was an active enforcer (e.g., the Brocade v. A10 Networks matters). However, I could not verify from the sources reached that US 7,995,580 itself was ever asserted in any of those campaigns. I will not impute the family's litigation history to this specific patent.

Two further practical points:

• The '580 issued August 9, 2011 and is recorded as Expired – Fee Related, with anticipated expiration May 15, 2021. Any assertion would necessarily post‑date the 2011 grant, and damages would be limited to pre‑expiration conduct.
• Brocade later became Avago Technologies International Sales Pte. Ltd. (2018 assignment). Avago‑side suits I located (e.g., ITC Inv. No. 337‑TA‑860; the IP Bridge/E.D. Tex. complaint against Avago) concern different patents (e.g., 6,947,456; 5,596,595) and cast Avago as a defendant, not as the '580 plaintiff.

3. Case table

Conclusion: no known litigation involving US patent 7,995,580.

4. Limitations of this search (stated honestly)

• I did not have live, authenticated access to PACER, the CAFC docket, or a commercial docket product (Docket Navigator / Lex Machina); I reached Justia, CourtListener/RECAP, Docket Alarm, PTABCTS, RPX and Unified Patents only through general web results. So I cannot certify the negative absolutely — only that no 7,995,580 case surfaced in any source reached.
• The CAFC/2026‑docket question was addressed in the earlier section; nothing new appeared here.

If you need a belt‑and‑suspenders confirmation, the highest‑yield free checks are: Judyrecords (searches case text and 1976‑2021 patents, per its documentation), the Unified Patents litigation caselist (https://portal.unifiedpatents.com/litigation/caselist) filtered to Foundry/Brocade/Avago, and a PACER/RECAP "party name" search on Brocade Communications, Foundry Networks, and Avago Technologies International Sales Pte. Ltd. for the 2011‑2021 window.

I'll verify the structured "no proceedings" result against live sources before writing up.

Proceedings overview

Zero. The structured PTAB-proceedings block (USPTO Open Data Portal, canonical source) returns no AIA trial proceedings on US 7,995,580 — 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution denials — so the defensive posture is not "hardened by surviving IPR" but rather "never tested at the Board at all, and now expired." That is a different, and for a defendant today largely favorable, posture: no proceeding exists because the patent's commercial relevance appears to have ended (anticipated expiration 2021-05-15, status "Expired – Fee Related"), not because challengers tried and failed.

I re-ran the search against live sources to make sure the ODP ingest wasn't simply lagging. Nothing surfaced. Details and the false positives I eliminated are below.

Proceedings on file

None — there are no AIA trial proceedings to profile

Because the requested per-proceeding template (petitioner, panel, grounds, institution decision, FWD, settlement, appeal) presupposes a docket, I am not going to populate it with invented case numbers. Instead, here is what I verified and how:

False positives I ruled out (important — these are not this patent). Searches on the string "558" produced a lot of noise from differently numbered patents whose opinions and Board papers use "'558" as shorthand:

• IPR2018-00255 — Quanergy Systems v. Velodyne Lidar, U.S. Patent 7,969,558 (lidar), FWD + rehearing denial: http://bannerwitcoff.com/wp-content/uploads/2020/06/PTAB-IPR2018-00255-62.pdf
• IPR2014-01555 / -01556 / -01557 / -01558 — Volkswagen v. EmeraChem Holdings, patents 5,451,558, 5,953,911, 7,951,346, 5,599,758 (catalyst/absorber).
• ITC/CAFC envelope-tracking materials (incl. the nonprecedential CAFC No. 2022-1046, Intel Corp. v. Qualcomm Inc., slip op. 2023-06-27, https://cafc.uscourts.gov/opinions-orders/22-1046.OPINION.6-27-2023_2148510.pdf) also use "'558 Patent" for a handset power-supply patent. Subject matter is facially unrelated to a backplane interface adapter, so this is a different patent — I could not resolve its full number from the snippets available to me, and I am flagging it rather than guessing.

None of these involve the Foundry/Avago 7,995,580 (backplane interface adapter with error control and redundant fabric). The number is unique and the authoritative full text confirms the assignment of US 7,995,580 B2 to Foundry backplane subject matter.

Why there is likely no PTAB history — the procedural ceiling this patent faced:

• PGR was never available. The '580 claims priority to 2000-11-17 and was filed 2009-03-09 — a pre-AIA patent. PGR under 35 U.S.C. § 321 is limited to first-inventor-to-file patents (effective filing date on/after 2013-03-16). Not applicable.
• CBM was theoretically available 2012–2020 but practically a non-starter. The '580 is a network-switch hardware patent — a classic "technological invention" under AIA § 18(d)(1), and CBM review sunset on 2020-09-16 anyway.
• IPR was the only realistic vehicle, and the window overlapped a period (2011–2021) when the patent was already owned by Brocade/Avago in a well-defended portfolio. No IPR was ever filed against it on the record available to me.

Strategic summary

Claim status: everything is UNTESTED. Every claim of 7,995,580 — independent and dependent alike — stands exactly as issued on 2011-08-09. No claim has been canceled, confirmed, or construed by the Board. (Note the cross-section caveat: the authoritative text supplied to me truncates before the claims, and I could not retrieve the '580 claim set verbatim, so I cannot give you a claim count or the claim numbers. Nothing in the record suggests any claim was ever amended by reissue or sua sponte — but confirm the claim listing in the Patent Center "Claims" tab for application 12/400,594 before relying on claim numbers in a filing.)

Estoppel landscape: no estoppel exists. 35 U.S.C. § 315(e)(2) estoppel attaches only after a final written decision in an IPR, and § 325(e)(2) after a PGR FWD. With no instituted proceeding and no FWD, no IPR/PGR estoppel bars anyone — petitioner, real party in interest, or privy. A defendant today has the complete § 102/§ 103/§ 112 toolbox available, both in district court and (subject to the note below) at the Board. There is also no de facto estoppel from a denied institution, because a denial of institution is non-appealable and creates no preclusive effect (see Cuozzo line of authority discussed in the 2016 OSB summaries surfaced in my search).

Family context, carefully bounded. The broader Foundry "high-performance network switch" family was aggressively challenged — but by A10 Networks in the Brocade v. A10 litigation (N.D. Cal. CV10-03428), and through reexamination, not AIA trials. The granted reexamination requests I saw cover US 7,774,833; 7,454,500; 7,899,899; 7,754,965; 7,647,427; 7,716,370 — including inter partes reexamination 95/001,803 (US 7,647,427) and ex parte reexamination 90/011,773. Those are pre-AIA reexaminations of sibling patents; they are not AIA trials and none of them is the '580. I also did not find the '580 among the asserted-patent lists in the Brocade v. A10 pleadings I reviewed — so I cannot tell you whether the '580 was ever asserted in litigation at all. Treat that as unverified.

Pattern signals: no repeat petitioner, no PTAB appeal history, no defensive aggregator (Unified Patents) in the chain for this number. That is the fingerprint of a patent that was owned and used by a large operating company in litigation (or not used at all), rather than one that was farmed to NPEs and hammered by IPRs.

Recommended next steps

1. If you are a defendant receiving a demand letter citing 7,995,580, the first question is not invalidity — it is standing and collectability. The patent is recorded as Expired – Fee Related, with anticipated expiration 2021-05-15 ("Anticipated expiration" event, 2021-05-15, per the Google Patents legal-status timeline at https://patents.google.com/patent/US7995580/en). Verify the maintenance-fee record and any revival petitions in USPTO Patent Center for application 12/400,594 — a lapsed patent is revivable only by petition, and an unrevived lapse undercuts the demand entirely.
2. Map the recoverable damages window. Under 35 U.S.C. § 286, a complaint filed today (2026-09-28) reaches back only to 2020-09-28. That means the entire actionable universe is roughly 2020-09-28 through 2021-05-15 (about 7½ months), plus any pre-2015 conduct for which a narrower bar applies. If the accused product/feature launched after 2021-05-15, there is no exposure at all.
3. Do not build a defense around an existing FWD, because there isn't one. There is no final written decision to quote and no canceled claim to point to. Any invalidity position must be built from scratch in the district court (§§ 102/103/112) or in a new Board proceeding. If you do consider an IPR: (a) an expired patent can still be the subject of an IPR, but the Board would apply Phillips claim construction rather than the broadest reasonable interpretation, and (b) the value is limited given the seven-month damages window — the ROI is usually poor.
4. Preserve the § 315(b) clock. If you have been served with a complaint alleging infringement of the '580, your IPR petition is due within one year of service. Calendar it even if you expect to rely on expiration.
5. Before filing anything, pull the verbatim claims. The authoritative full text I was given truncates before the claims, and I did not verify the '580 claim set. Get it from the granted-patent PDF (https://patents.google.com/patent/US7995580/en, "Claims") or USPTO Patent Center for 12/400,594. Also confirm the '580's terminal-disclaimer status relative to US 7,596,139 and US 7,236,490 — sibling members of this family carry terminal disclaimers, and that affects both enforceability and how much term ever existed.

Uncertainty statement: the "no PTAB activity" conclusion is sourced from the canonical USPTO ODP block and corroborated by multiple public sources, but I did not have live interactive access to PTAB E2E / Patent Center docketing or to Docket Alarm's PTAB index; my confirmation is search-based. If an IPR was filed and terminated very early (e.g., a pre-institution settlement or a denied petition that never indexed), it could have escaped all of the sources I reached. Re-verify against PTAB E2E (https://e2e.uspto.gov) and the PTAB Decisions page before you certify "no IPR" in a client memo or a court filing.

I'll research the assignment record for US 7,995,580 through the available sources.

Let me look at the specific assignment data sources.

Let me search for the specific security agreement and transfer records.

I've hit the tool-call limit, so I'll now compose the answer from the authoritative patent text (which embeds Google Patents' legal-events data for the '580) plus the assignment corroboration surfaced in my searches. Where I could not reach live Assignment Center records, I say so explicitly rather than reconstruct.

Ownership Chain — US 7,995,580 B2

Verification note up front: The live USPTO Patent Assignment Search (https://assignmentcenter.uspto.gov/, mirrored at https://assignment.uspto.gov/patent/index.html) was not reachable from the tooling available to me, so I could not pull the per-patent assignment abstract, the correspondent of record, or the reel/frame for every link directly. What follows is built from (a) the Google Patents Legal Events block embedded in the authoritative full text supplied for this patent, and (b) corroborating reel/frame data that surfaced on sibling Brocade/Foundry patents in the same portfolio-wide security recordings. I flag every point where a reel/frame or correspondent is unconfirmed for the '580 specifically.

Inventors

Employer basis: The inventors' rights were assigned to Foundry Networks, Inc. (recorded 2011‑01‑28, conveyance Assignment of Assignors' Interest, assignors WONG / CHANG / LIN / PATEL / WONG). The 2002 parent publication listed only four inventors; Yuen Fai Wong was added later (consistent with the current grant).

Anomaly to flag (data-level): The assignment record's assignor string reads "WONG, ARIS, CHANG, ANDREW, LIN, YU‑MEI, PATEL, RONAK, WONG, MING G." The token "WONG, ARIS" does not match any named inventor (the two Wongs are Ming G. Wong and Yuen Fai Wong). This is most likely an OCR/transcription artifact in the assignment index rather than a sixth assignor, but I cannot confirm it against the live record. Flagging it as recorded, not correcting it.

Departure pattern: No evidence of the "all inventors leave within 12 months" fire-sale precursor. The change in ownership was driven by corporate M&A (Brocade's acquisition of Foundry in 2008), not by inventor attrition. I could not determine individual inventor departure dates — not verified.

Original assignee

• Entity on the issued patent: Foundry Networks, Inc. (San Jose, CA). (FreepatentsOnline lists the assignee of record as "Foundry Networks, Inc.")
• Primary line of business: High-performance Layer 2/3/4‑7 Ethernet switches and routers (the BigIron, FastIron, ServerIron, NetIron families) — a direct competitor of Cisco in the enterprise/service-provider switch market.
• Did they ship a product embodying the claims? Yes, with high confidence at the disclosure level: the '580 is the backplane-interface-adapter/redundant-fabric continuation out of the same Foundry "high-performance network switch" program (the 5 × 10 Gbps striped-backplane architecture) that produced the shipped Foundry chassis products; the specification repeatedly frames the cell-size choice around 4 × 10 GE line rate through the backplane. (I did not independently verify a specific SKU against the claims, so treat "embodies" as disclosure‑grounded.)
• Current status: No longer independent. Foundry Networks, Inc. was acquired by Brocade Communications Systems (deal closed Dec 2008). The assignment record shows Foundry Networks, Inc. → Foundry Networks, LLC by Change of Name (recorded 2010‑07‑21). Brocade itself was acquired by Broadcom (Nov 2017), and the Foundry/Brocade patent estate (including the '580) was conveyed to Avago Technologies International Sales Pte. Ltd. on 2018‑10‑18. So the original assignee is now a renamed subsidiary shell inside Broadcom, not a going-concern brand.

Assignment timeline

Reel/frame entries marked [sibling-confirmed] were surfaced on other Brocade/Foundry patents in the same portfolio-wide recording (e.g., US 2010/0049999 legal events) and therefore carry the same reel/frame; I could not open the '580-specific abstract to read the frame back for this patent. Entries marked [reel unverified] had no reel/frame in the sources reached.

• 2010‑01‑20 (executed) / recorded 2010‑01‑20 — Reel 023814/0587 [sibling-confirmed]

  • Conveyance: Security Agreement (grant of security interest)
  • Assignor: Brocade Communications Systems, Inc.; Foundry Networks, LLC; Inrange Technologies Corporation; McDATA Corporation; McDATA Services Corporation (and others)
  • Assignee: Wells Fargo Bank, National Association, as Collateral Agent
  • Correspondent: not retrieved (live record not reachable)
  • Context: Securitization — collateral pledge securing Brocade's 6.625% 2018 and 6.875% 2020 Senior Secured Notes (confirmed via the June 18, 2010 Cooley LLP Exhibit 5.1 registration opinion for the S‑4 covering the same guarantor group).

• 2010‑07‑21 (executed) / recorded 2010‑07‑21 — Reel [unverified]

  • Conveyance: Change of Name
  • Assignor: Foundry Networks, Inc.
  • Assignee: Foundry Networks, LLC
  • Correspondent: not retrieved
  • Context: Internal reorg / change of name only — entity conversion within the Brocade group; no ownership change of substance.

• 2011‑01‑28 (executed) / recorded 2011‑01‑28 — Reel [unverified]

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Wong (as "ARIS"), Chang, Lin, Patel, Wong
  • Assignee: Foundry Networks, Inc.
  • Correspondent: not retrieved
  • Context: Confirmatory inventor→company assignment — the inventors' employment assignment, recorded ~10 years after the 2001 filing and around the grant date, evidently to perfect the record chain.

• 2011‑09‑20 (executed) / recorded 2011‑09‑20 — Reel [unverified]

  • Conveyance: Supplemental Patent Security Agreement
  • Assignor: Brocade Communications Systems, Inc.; Foundry Networks, LLC; McDATA Corporation
  • Assignee: Bank of America, N.A., as Administrative Agent
  • Correspondent: not retrieved
  • Context: Securitization — supplemental collateral pledge to a second secured lender.

• 2011‑09‑26 (executed ~2011‑09‑16) / recorded 2011‑09‑26 — Reel 026971/0042 [sibling-confirmed]

  • Conveyance: Supplemental Patent Security Agreement
  • Assignor: Brocade; Foundry Networks, LLC; Inrange; McDATA; McDATA Services
  • Assignee: Wells Fargo Bank, National Association, as Collateral Agent
  • Correspondent: not retrieved
  • Context: Securitization — supplemental pledge under the same Wells Fargo collateral agency.

• 2015‑01‑21 (executed ~2015‑01‑14) / recorded 2015‑01‑21 — Reel [unverified]

  • Conveyance: Release by Secured Party
  • Assignor: Bank of America, N.A., as Administrative Agent
  • Assignee: Brocade Communications Systems, Inc.; Foundry Networks, LLC
  • Correspondent: not retrieved
  • Context: Release of security interest — debt retired, collateral lien lifted.

• 2015‑01‑22 (executed ~2015‑01‑14) / recorded 2015‑01‑22 — Reel 034804/0793 [sibling-confirmed]

  • Conveyance: Release of Security Interest
  • Assignor: Wells Fargo Bank, National Association, as Collateral Agent
  • Assignee: Foundry Networks, LLC; Brocade Communications Systems, Inc.
  • Correspondent: not retrieved
  • Context: Release of security interest — parallel lien release with the BofA release above.

• 2018‑10‑18 (executed) / recorded 2018‑10‑18 — Reel [unverified]

  • Conveyance: Assignment of Assignor's Interest
  • Assignor: Brocade Communications Systems LLC
  • Assignee: Avago Technologies International Sales Pte. Ltd.
  • Correspondent: not retrieved
  • Context: Acquisition / M&A IP transfer — Broadcom (parent of Avago Technologies International Sales) absorbed the Brocade IP estate following the 2017 Broadcom–Brocade acquisition.

If the Assignment Center returns records inconsistent with the above (in particular the '580-specific reel/frame and the correspondents), prefer those live records — I could not open them.

Repeat-correspondent signal: cannot be assessed. The task's key tell — a single repeat-player attorney filing every recording for an NPE family — could not be tested because no correspondent of record was retrievable for any link. I will not invent one.

Timeline diagram

timeline
    title Ownership of US 7995580
    2000 : Priority filing by inventors
    2001 : Application filed
    2008 : Foundry acquired by Brocade
    2010 : Wells Fargo security agreement
         : Foundry renamed to LLC
    2011 : Inventor assignment recorded
         : Patent issued
         : Supplemental security agreements
    2015 : Security interests released
    2018 : Assigned to Avago Technologies
    2021 : Anticipated expiration

NPE / troll-pattern signals

Verdict

Operating-company assertion — with an express caveat that the "assertion" element is unverified for this specific patent.

Justification: The entire recorded chain runs through operating companies — original assignee Foundry Networks, Inc. (a switch vendor shipping the BigIron/FastIron/NetIron line), renamed Foundry Networks, LLC (Reel for 2010‑07‑21), whose assets passed via Brocade to Avago Technologies International Sales Pte. Ltd. (Reel for 2018‑10‑18). There is no shell-entity transfer, no known asserter, no cascading LLC chain, no bankruptcy sale, and no defensive-aggregator terminus; the intervening recordings are securitizations and releases (Reels 023814/0587, 026971/0042, 034804/0793) and a change of name. Zero of the eight NPE signals are present on the evidence reached. The only reason I do not call this a clean "operating company suing competitors" is that no infringement suit naming the '580 was located in the prior section, and the correspondent-of-record test (Signal 3) could not be run at all because the live Assignment Center records were unreachable.

To close the two open items, run these against the live systems (which I could not reach):

• USPTO Patent Assignment Search — patent number 7995580: https://assignmentcenter.uspto.gov/ (mirror: https://assignment.uspto.gov/patent/index.html) — to capture the '580-specific reel/frame and the correspondent of record for the 2010‑07‑21, 2011‑01‑28, 2011‑09‑20, 2015‑01‑21, and 2018‑10‑18 links, and to test the repeat-correspondent signal.
• USPTO PEDS / ODP for application 12/400,594, and SEC EDGAR for the Brocade S‑4 (CIK 1009626) and the 2017 Broadcom–Brocade merger filings, to confirm the chain-of-title narrative behind the 2010 securitizations and the 2018 Avago transfer.

Two flagged contradictions/uncertainties to carry forward: (1) the assignor token "WONG, ARIS" in the 2011‑01‑28 record matches no named inventor; and (2) every reel/frame above except the three Wells Fargo entries is labeled [unverified] because I could not open the '580-specific record — do not treat those reel/frames as confirmed.

I'll start by searching for the USPTO record and citation list for US 7,995,580.

Let me pull the cited-references list for the '580 specifically from uspto.report and other sources.

Prior-art analysis — US 7,995,580 B2

0. How this was obtained, and the two hard limits on it

What I searched. I searched for the literal string 7995580 and for the '580's reference list. I did not return or attribute anything for similar-looking numbers (e.g., 9,039,580 / Woodway treadmill, or the PACER receipt number "AVAEDC‑7995580" identified in the earlier litigation section) — consistent with the strict "no auto-correction" rule.

Limit 1 — no direct USPTO database session. I could not open USPTO Patent Center / Patent Full-Text (patft) directly. The citation list below was reconstructed from (a) the uspto.report mirror of the USPTO front page for grant 7995580 (https://uspto.report/patent/grant/7995580), and (b) the front page of the family sibling US 9,030,937 B2 PDF (same title, same inventors, continuation of this very application 12/400,594), which lists the same class of "References Cited." I flag where the two may differ.

Limit 2 — I still do not have the '580's verbatim claims. The authoritative full text supplied to me is truncated before the claims, exactly as noted in the earlier summary section. I will therefore not fabricate claim numbers. In §4 I map art to claim subject matter as disclosed, and where a specific claim number is asserted it is labeled unverified. This is the single biggest caveat in this memo, and I am flagging it rather than papering over it.

One process note on the record: the earlier summary lists the '580's examiners as Primary Examiner Chi H. Pham / Assistant Examiner Kevin Mew. The sibling '937 front page shows "Primary Examiner — Kevin Mew" and attorney Fountainhead Law Group P.C. That is consistent with Mew having been promoted between 2011 and 2015, not a contradiction — but the '580's own examiner assignment should be read off the granted front page.

Statutory framework (important for this patent). The '580 was filed 9 March 2009, i.e., pre‑AIA §102 applies. Its effective filing date is at the earliest 17 Nov 2000 (provisional 60/249,871) / 15 May 2001 ('038 parent, now 7,236,490) / 16 Nov 2001 ('066 parent, now 7,596,139). The '937 continuity data supplied in the earlier record confirms: "'580 … continuation of 09/988,066 … now Pat. No. 7,596,139, and a continuation‑in‑part of 09/855,038 … now Pat. No. 7,236,490."

Consequence: almost the entire cited list post-dates the invention and can only be prior art under pre‑AIA §102(e) (a U.S. patent granted on an application filed before the applicant's effective filing date), not §102(a)/(b). Only references published before ~17 Nov 1999 / 15 May 2000 can be §102(b) art. That structural point governs the whole table below.

1. The '580's cited references — as listed

The uspto.report rendering of the '580 front page gives the U.S. Patent Documents in ascending numeric order. Representative entries (number | issue date | inventor(s) as listed):

6147996 | 11/2000 | Laor et al. · 6154446 | 11/2000 | Kadambi et al. · 6157643 | 12/2000 | Ma · 6160809 | 12/2000 | Adiletta et al. · 6172990 | 01/2001 | Deb et al. · 6178520 | 01/2001 | DeKoning et al. · 6181699 | 01/2001 | Crinion et al. · 6222845 | 04/2001 | Shue et al. · 6243667 | 06/2001 | Kerr et al. · 6263374 | 07/2001 | Olnowich et al. · 6272144 | 08/2001 | Berenbaum et al. · 6304903 | 10/2001 | Ward · 6320859 | 11/2001 | Momirov · 6333929 | 12/2001 | Drottar et al. · 6335932 | 01/2002 | Kadambi et al. · 6335935 | 01/2002 | Kadambi et al. · 6343072 | 01/2002 | Bechtolsheim et al. · 6351143 | 02/2002 | Guccione et al. · 6356550 | 03/2002 | Williams · 6356942 | 03/2002 | Bengtsson et al. · 6363077 | 03/2002 | Wong et al. · 6369855 | 04/2002 | Chauvel et al. · 6421352 | 07/2002 | Manaka et al. · 6424658 | 07/2002 | Mathur · 6424659 | 07/2002 | Viswanadham et al. · 6427185 | 07/2002 | Ryals et al. · 6460088 | 10/2002 | Merchant · 6463063 | 10/2002 | Bianchini, Jr. et al. · 6466608 | 10/2002 | Hong et al. · 6470436 | 10/2002 | Croft et al. · 6473433 | 10/2002 | Bianchini et al. · 6477174 | 11/2002 | Dooley et al. · 6480477 | 11/2002 | Treadaway et al. · 6490280 | 12/2002 | Leung · 6493347 | 12/2002 | Sindhu et al. · 6496502 | 12/2002 | Fite et al. · 6522656 | 02/2003 | Gridley · 6532229 | 03/2003 | Johnson et al. · 6532234 | 03/2003 | Yoshikawa et al. · 6535504 | 03/2003 | Johnson et al. · 6549519 | 04/2003 | Michels et al. · 6553370 | 04/2003 | Andreev et al. · 6556208 | 04/2003 | Congdon et al. · 6567404 | 05/2003 | Wilford · 6587432 | 07/2003 | Putzolu et al. · 6591302 | 07/2003 | Boucher et al. · 6654342 | 11/2003 | Dittia et al. · 6654346 | 11/2003 | Mahalingaiah et al. · 6654370 | 11/2003 | Quirke et al. · 6654373 | 11/2003 | Maher, III et al. · 6658002 | 12/2003 | Ross et al. · 6661791 | 12/2003 | Brown · 6671275 | 12/2003 | Wong et al. · 6678248 | 01/2004 | Haddock et al. · 6681332 | 01/2004 | Byrne et al. · 6687247 | 02/2004 | Wilford et al. · 6691202 | 02/2004 | Vasquez et al. · 6696917 | 02/2004 | Heitner et al. · 6697359 | 02/2004 | George · 6697368 | 02/2004 | Chang et al. · 6700894 | 03/2004 | Shung · 6708000 | 03/2004 | Nishi et al. · 6721229 | 04/2004 | Cole · 6721268 | 04/2004 | Ohira et al. · 6721313 | 04/2004 | Van Duyne · 6721338 | 04/2004 | Sato · 6735218 | 05/2004 | Chang et al. · 6745277 | 06/2004 | Lee et al. · 6751224 | 06/2004 | Parruck et al. · 6754881 | 06/2004 | Kuhlmann et al. · 6765866 | 07/2004 | Wyatt · 6775706 | 08/2004 | Fukumoto et al. · 6778546 | 08/2004 | Epps et al. · 6781990 | 08/2004 | Puri et al. · 6792484 | 09/2004 | Hook · 6792502 | 09/2004 | Pandya et al. · 6798740 | 09/2004 | Senevirathne et al. · 6804220 | 10/2004 | Odenwalder et al. · 6804731 | 10/2004 | Chang et al. · 6807179 | 10/2004 | Kanuri et al. · 6807363 | 10/2004 | Abiko et al. · 6810046 | 10/2004 | Abbas et al. · 6813243 | 11/2004 | Epps et al. · 6813266 | 11/2004 | Chiang et al. · 6816467 | 11/2004 | Muller et al. · 6831923 | 12/2004 | Laor et al. · 6831932 | 12/2004 | Boyle et al. · 6836808 | 12/2004 | Bunce et al. · 6839346 | 01/2005 | Kametani · 6842422 | 01/2005 | Bianchini · 6854117 | 02/2005 | Roberts · 6856600 | 02/2005 | Russell et al. · 6859438 | 02/2005 | … (list truncated at the point my fetch ended)

The sibling '937 front page additionally shows the classic older citations typical of this family, e.g. 3,866,175 | 2/1975 | Seifert, Jr. et al., 4,628,480 | 12/1986 | Floyd, 4,667,323 | 5/1987 | Engdahl et al., 4,683,564 | 7/1987 | Young et al., 4,698,748 | 10/1987 | Jurswik et al., 4,723,243 | 2/1988 | Joshi et al., 4,754,482 | 6/1988 | Weiss, 4,791,629 | 12/1988 | Burns et al., 4,807,280 | 2/1989 | Posner et al., 4,876,681 | 10/1989 | Hagiwara et al., 4,896,277 | 1/1990 | Vercellotti et al., 4,985,889 | 1/1991 | Franzish et al., 5,101,40… (truncated). Foreign patent documents on the '937 front page: EP 1380127 A2 | 1/2004 and JP 2003/289359 A | 10/2003. Other publications: the family's own prosecution papers (e.g., "Non‑Final Office Action for U.S. Appl. No. 13/548,116 mailed on Apr. 15, 2013, 8 pages").

⚠️ Do not treat the list as exhaustive or as photo-verified. The uspto.report fetch ended mid‑list, and I could not confirm whether the '580's own front page includes foreign documents and NPL in the same form as the '937. Any §102 conclusion of the form "everything cited" needs a look at the granted PDF front page.

Also note: Google Patents' "Prior art keywords" for the '580 are data, cell, error, packet, synchronization, and the classifications are H04L49/153 (ATM switching fabrics having parallel switch planes), H04L49/1538 (cell slicing), H04L49/25 (routing in a switch fabric), H04L49/3063 (pipelined peripheral units), H04L49/901 (buffering using storage descriptors), H04L47/6225 (round-robin fixed service order), H04L49/352 (Gigabit Ethernet switching). That tag set tells you where the examiner was looking: parallel-plane/cell-sliced fabrics, in-band cell control, and queue arbitration.

2. Why "self/own-family" citations matter here more than the ordinary art

Several cited references are the applicant's own family and inventor-group art: 6,697,368 Chang et al., 6,735,218 Chang et al., 6,804,731 Chang et al., 6,771,275 Wong et al., plus (as background/related) 7,236,490 Chang et al. "Backplane interface adapter", 7,512,127 Chang et al. "Backplane interface adapter", 7,206,283 Chang et al. "High-performance network switch", 7,203,194 Chang et al. "Method and system for encoding wide striped cells", 7,278,702/7,356,030 "Network switch cross point".

This is the single most legally consequential point in the whole citation set. Because the '580 is, per its own continuity data, a continuation of 09/988,066 and a continuation‑in‑part of 09/855,038, any claim of the '580 that is not supported by the earlier 2000/2001 disclosure (i.e., any claim to the added "error control" and "redundant fabric" subject matter that gives the patent its title) loses the earlier priority date for that claim. Those claims then face:

• US 7,236,490 (Chang et al., issued 26 Jun 2007) as potential §102(b) art (issued >1 yr before the 9 Mar 2009 filing);
• US 2002/0105966 A1 (the '066 publication, published 8 Aug 2002) and US 7,596,139 as §102(e) art; and
• the Chang cross-point/wide-cell patents as §102(e) art.

This is exactly how a CIP child gets invalidated, and the examiner's citation of the Chang wide-cell/fabric patents is consistent with that awareness. It is also, per the earlier sections, the sort of issue that never surfaced in litigation because no case asserting the '580 was found.

3. Tier‑1 references — the ones that actually matter

I give descriptions only where I have real confidence; where I don't, I say so rather than invent a title. Any bracketed "[title/subject not verified in this session]" means precisely that.

Two more categories in the list deserve naming even without itemized titles:

• Classic pre‑1990 backplane/bus art (3,866,175, 4,628,480, 4,667,323, 4,683,564, 4,698,748, 4,723,243, 4,754,482, 4,791,629, 4,807,280, 4,876,681, 4,896,277, 4,985,889, …). These are §102(b) art (published well over one year before any possible priority date). They are almost certainly cited for only one purpose: to show that computing bus/backplane interfaces, FIFOs and synchronization are ancient, i.e., to support §103 obviousness arguments and to constrain the scope of "interface adapter." I cannot describe each individually with confidence, and I will not invent characterizations.
• Foreign documents EP 1380127 A2 (Jan 2004) and JP 2003/289359 A (Oct 2003) — both post‑date the invention, so they can only be pre‑AIA §102(e)-equivalent / §102(a)/(b)-art by their earlier filing or priority dates, or be family members (the JP family member JP2004537871A was identified in the earlier summary as having 137 claims). If they are family members they are not prior art at all. This needs checking on the front page.

4. §102 mapping by claim theme — with the claim-number caveat stated

Caveat (repeat): the '580's literal claims are not in my possession. The "claim(s)" column below refers to claim categories as disclosed and as expressly framed as inventive in the specification, not to verified claim numbers. Do not cite a claim number from this memo in a validity contention.

§102(b) versus §102(e), plainly: on this record, essentially none of the 2002–2005‑issued U.S. references is §102(b) art against a claim with 2000/2001 priority. They are §102(e) art at best. The only true §102(b) references in the set are (i) the pre‑1990 bus/backplane patents, (ii) the pre‑Nov‑1999 published material, and (iii) US 7,236,490 as against new‑matter claims. That asymmetry is the most important practical takeaway of this whole analysis.

5. "Which claims does each reference anticipate?" — the honest answer

Strictly: I cannot give a reference‑by‑reference "anticipates claim X" answer, because (a) I do not have the '580's claim text, and (b) anticipation requires an element‑by‑element comparison against that text. What I can state with confidence, and have stated above, is:

1. The cited set is overwhelmingly §102(e)-only art, not §102(b) art — because the '580's effective filing date is 2000/2001 and the cited patents issued 2002–2005.
2. Bianchini '422/'433/'063 is the closest cited art on the "redundant fabric" subject, and Olnowich '374 the closest cited art on the "interface adapter" subject.
3. The striping + in‑band synchronization character + queue‑level synchronization‑error detection combination — the true core of the '580 — has no clearly anticipatory reference in the list I was able to retrieve, which is consistent with this patent having been allowed over that art.
4. The most dangerous reference is the applicant's own parent, US 7,236,490 (and the '066/'139 parents), under the CIP‑new‑matter theory in §2 — not any third‑party reference.

6. Verification steps that would close the gaps

1. Granted PDF front page (https://patents.google.com/patent/US7995580/en → "References Cited") for the definitive, complete "U.S. Patent Documents," "Foreign Patent Documents" and "Other Publications" lists, plus the examiner names.
2. USPTO Patent Center, application 12/400,594 → the IDS/PTO‑1449 forms and the examiner's Form 892, which give the category codes and the examiner's own field‑of‑search and art groupings — i.e., the examiner's actual §102/§103 theories, which is what the question is really asking for. This is not obtainable through general web search.
3. The '580 claim set (Claims section of the granted PDF) — indispensable before any claim‑numbered §102 statement.
4. Filing dates (not just issue dates) for each cited reference, to confirm which ones are within pre‑AIA §102(e) as of 17 Nov 2000 / 15 May 2001 / 16 Nov 2001.
5. For the two foreign documents, confirm whether they are family members (no prior‑art effect) or true third‑party art.

Flagged discrepancy for the record: Unified Patents' page for US‑20090290499‑A1 (the '645 sibling) reports priority date 2000‑11‑16 and application date 2009‑03‑08, whereas Google Patents reports 2000‑11‑17 / 2009‑03‑09 for 12/400,594. This does not change any §102 conclusion, but the one‑day offsets indicate third‑party date drift that should be resolved from the USPTO record before relying on any critical date.

I'll first try to retrieve the actual Prior Art section (cited references) and claim text for US 7,995,580, since the authoritative full text supplied to me is truncated before the claims and does not include the citations list.

I have partial prior-art lists now. Let me verify the key references and try once more for the claim text.

Obviousness analysis of US 7,995,580 under 35 U.S.C. § 103

Important update to the earlier sections of this analysis. The prior sections stated the '580's claims could not be retrieved and declined to discuss claim language. In this pass I retrieved what is presented as the granted claim set (claims 1–11) on a third‑party mirror of the printed patent, together with the "References Cited" listing. That changes the analysis materially, so I flag the update rather than restating the earlier caveat. Everything below is keyed to that text; where a reference's disclosure (as opposed to its title/number/date) is being inferred rather than verified, I say so.

1. Governing law and the critical date (this matters more than usual here)

Because application 12/400,594 was filed March 9, 2009, pre‑AIA § 103(a) applies (pre‑March 16, 2013 filing), and pre‑AIA § 103(c) is available to disqualify commonly owned art.

The '580 is a continuation of 09/988,066 (now US 7,596,139), filed Nov. 16, 2001, which is itself a CIP of 09/855,038 (US 7,236,490) with benefit of provisional 60/249,871 (Nov. 17, 2000) (continuity data). The error‑control subject matter that the '580 claims is described in 09/988,066 (published as US 2002/0105966 A1), so the claims are presumptively entitled to Nov. 16, 2001.

Three consequences that dominate the analysis:

1. The patent's own "References Cited" list is not a prior‑art list for these claims. It was submitted in the 2009 continuation, and much of it post‑dates Nov. 16, 2001 (e.g., 6,798,740 granted 2004; 6,901,072 granted 2005; 6,721,313 granted 2004). Such documents are § 103 art only if their own U.S. filing dates pre‑date the invention (pre‑AIA § 102(e), or § 102(e)/(g) via the patentably‑identical doctrine). Any § 103 ground built on the face‑of‑patent list must survive that date check. This is the single most common error in invalidating a late‑filed continuation.
2. Family members are not art. US 6,697,368, 6,735,218, 7,236,490, 7,596,139, 7,948,872, 8,514,716, 8,619,781, 8,964,754 and 9,030,937 share the same inventive entity (Patel/Wong/Lin/Chang/Wong) and/or were commonly owned by Foundry at the time of invention; they are disqualified for § 103 by § 103(c) (or fail "by another"). They appear in the record as IDS references only.
3. The claims do not require striping, wide cells, or redundant fabric. This is the most important substantive observation: despite the title, "error control and redundant fabric," the retrieved claims are directed only to the error‑control feedback loop. The redundant‑fabric subject matter (RFTs, dual switching blades) is unclaimed. So a § 103 attack should be aimed at link/FIFO health monitoring + back‑pressure signaling to the source, not at redundancy art. (Note the earlier sections' four bulleted "inventive" concepts — only #1–#3 are reflected in these claims.)

2. What the claims actually require

Retrieved claim text (uspto.report mirror for the '580 grant) — claim 1 and the two other independent claims, abridged:

Caveat (do not skip). This text came from a single third‑party mirror; Google Patents' own claims section was not returned by the tools. The claim set is internally consistent with the '580's abstract, parent‑case text and Reference list, but for any invalidity contention the literal language must be pulled from the grant PDF or Patent Center. I also note, as an aside outside § 103, that claim 2 and 5 refer to "said level monitor"/"said synchronization error detector" with no antecedent in claim 1, and claim 1 introduces a "flow controller" without reciting it as part of the device — potential § 112(b) issues that a challenger would also plead.

3. The prior‑art universe available from this page

The page's "prior art keywords" are data, cell, error, packet, synchronization, and the prior‑art date is listed as 2000‑11‑17 — i.e., the Office/Google tags this family as error‑ and synchronization‑centric packet‑switch art, which is exactly where the § 103 inquiry belongs.

From the References Cited listing (uspto.report) and the sibling 7236490 FPO listing (which gives titles), the references whose titles I could verify and that are pertinent to claims 1–11 are:

Trap to avoid: EP 1 271 323 A2/A3, "Redundant backplane interconnect for electronic devices" (HP, deBlanc/Dickey/White) (EPO PDF). It looks perfect for the title's "redundant fabric," but its priority is 21 June 2001 and it published 2 Jan. 2003 — after Nov. 16, 2001. As a foreign (EP) publication it gets no § 102(e) benefit, so it is not prior art to these claims. Likewise, because the retrieved claims don't recite redundancy at all, redundancy art is largely beside the point.

4. Combination A — claims 1, 2, 6, 7, 8, 9

Suggested ground: Bianchini (US 6,842,422) or Van Duyne (US 6,721,313) in view of Senevirathne (US 6,798,740) and Quirke (US 6,654,370); further in view of the pre‑2001 switch‑ASIC references (e.g., Kadambi 6,154,446; Adiletta 6,160,809) for per‑flow queuing.

Element mapping (with confidence):

5. Combination B — claims 3, 4, 10, 11

Add: Blanc (US 6,606,300), "Flow control process for a switching system," and/or the pre‑2001 flow‑control art already in the record.

• Claim 3/10's "recovery routine to re‑synchronize" and claim 4's "throttling back the data flowing" are back‑pressure — the core of Blanc and of standard switch flow control. Once claim 1's error is detected, asserting back‑pressure (pause/stop sending until the buffer drains) and then re‑establishing a known sync state is the only conventional response; no new mechanism is required.
• Claim 11 ("flow controller sends the second signal to other components of the system to begin recovery measures") is ordinary broadcast/notification of a fault to peer components (cf. the '580's own AOP‑to‑all‑slots and "Fabric Active" signalling, and Senevirathne's system‑level response).

6. Combination C — claim 5 (the K2 character)

Add: the applicant‑admitted 10 GbE/8b‑10b art of record (10 Gigabit Ethernet – Technology Overview White Paper, Sept. 2001; Foundry BigIron/JetCore briefs; the Gigabit Ethernet Alliance and IEEE 802.1D materials cited in the '580's Other References).

• The specification itself concedes the characters are industry‑standard 8b/10b K‑codes it repurposes: K0 = start of cell, K1 = end of packet, K2 = stripe sync. XAUI/Fibre Channel/XAUI‑style links carried reserved/ordered sets, including comma/IDLE and sync ordered sets, precisely for link re‑alignment. Selecting the third available reserved K‑code as the sync character is a design choice among a finite, identified set — KSR/result‑effective‑variable territory (In re Boesch; Pfizer).
• Nothing in claim 5 recites what K2 is beyond being a character identified during recovery, so the claim is satisfied by any distinctive sync‑sequence character — including ordered sets already used for link re‑sync in the admitted standards art.

7. Motivation to combine — the substantive part

1. The problem is inherent in the architecture, not discovered by the inventors. Once you stripe traffic from one blade across multiple serial planes into per‑source receive queues (Bianchini; Van Duyne; the admitted Foundry briefs), a break in any single source/plane path necessarily manifests as one queue draining while its neighbours overrun. Any designer of such a system confronts exactly the § 1 problem statement. KSR permits "the inferences and creative steps" the artisan would employ, and "design incentives and other market forces" as motivation.
2. There are a finite number of predictable solutions. The artisan's toolbox in 2000–2001 was: (a) monitor link/element health (Senevirathne); (b) monitor buffer occupancy and use watermarks (Blanc; Sindhu; the pre‑2001 switch ASICs); (c) signal back‑pressure upstream to the traffic source (Blanc); (d) re‑align the parallel paths with a distinctive sync sequence (Quirke; the 8b/10b/10 GbE standards). Every element of claims 1–11 is one of these known moves, and their combination does no more than yield the predictable aggregate result (detect → notify → stop → re‑sync).
3. Cross‑citation and common field. The references are all in packet switching/backplane interconnect — same field of endeavor and reasonably pertinent to the same problem (Unwired Planet v. Google framing). Senevirathne and Blanc even flag their applicability to fabric cores and switch systems generally, and the '580's own IDS groups them together, evidence that those of skill regarded the art as one body (In re Nuvasive/KSR "same field" + "addressing the same technical issue" rationales).
4. Known technique, same way, predictable result. Monitoring the level of a receive FIFO to detect a stalled/failed source, and stopping the source, is the ordinary "known technique used in the same way" rationale; nothing in the claims reconfigures the elements so that they interact in an unexpected way.
5. Claim 1 is broad enough that even simple art suffices. Claim 1 requires only that a receiving blade counts/monitors incoming data, detects an error for the source from that amount, and causes a signal back to the source via a flow controller. It does not require striping, wide cells, FIFOs, K2, slot‑versus‑processor granularity, or any threshold. Any fabric with per‑link receive‑side error/occupancy detection and upstream fault signalling reads on it. This is the claim's chief liability and, correspondingly, the family's later continuations ('781, '716, '937) were drafted around stripe‑control‑interval and wide‑cell encoding features the '580 claims eschew (see, e.g., the '781 claim 1 as reported by RPX).

8. Where the § 103 case is weak, and the best rebuttals a patent owner would raise

1. No reference may disclose detecting the error from the amount of data monitored per source. This causal link ("based on the amount of data monitored," claim 1; overflow/underflow, claim 2; "traffic flow received and traffic synchronization," claim 10) is the natural point of attack back: the owner will argue the prior art detects link faults directly (SERDES error flags, /E characters, GLINK, cross‑point error wires — all described in the '580 spec), not by inferring a per‑source sync failure from queue occupancy. The counter is that deriving link/sync health from buffer occupancy is exactly what quotient watermark monitoring does, and the '580 spec itself treats it as an ordinary implementation ("can be detected by monitoring the FIFOs and detecting an overflow and/or underflow"), and simultaneously discloses the direct‑detection alternative — i.e., the inventors themselves present the two as equivalent options, undercutting any "unexpected" argument.
2. Claim 5's K2 and the 112‑count implementation. The claim is broad, but if the owner argues the 112‑K2 sequence matched to sync‑queue depth (a fully disclosed, non‑claimed detail) supplies the inventive weight, the answer is that unclaimed subject matter cannot supply patentability and the count is a result‑effective variable of buffer depth.
3. Secondary considerations. Likely owner arguments: commercial success of Foundry's JetCore/BigIron chassis, long‑felt need for 50 Gbps/slot non‑blocking backplanes, and copying. These are weak here because (a) the touted commercial advance in this family is the wide‑cell/160‑byte/line‑rate design, which the '580 claims do not cover; and (b) nexus would be hard to establish for a data‑path error‑feedback loop, since the same chassis switched traffic without regard to which error‑handling variant was in silicon. Note the family's own specification expressly disparages the conventional parallel‑bus approach and frames the serial link insight as the advance — again, unclaimed here.
4. § 103(c) / "by another." Any ground that leans on the Foundry family patents (6,697,368; 6,735,218; 6,901,072; 7,236,490; 7,596,139) fails as a matter of law. The strongest ground must use third‑party art with confirmed pre‑Nov. 16, 2001 filing dates.
5. Enablement/analogous‑art skirmishes. Senevirathne (switch‑core health/redundancy) may be argued non‑analogous by an owner, but it is same‑field (switch fabric health) and reasonably pertinent to the very problem of detecting a failed fabric element and notifying the system.

9. Conclusion on obviousness

• Claims 1, 3, 6, 7, 8, 9, 10 and 11 present a strong prima facie § 103 case as a combination of (i) a serial/striped multi‑plane backplane switching system (Bianchini 6,842,422; Van Duyne 6,721,313; George 6,697,359), (ii) switch‑core/link health monitoring with system notification (Senevirathne 6,798,740; Quirke 6,654,370), and (iii) switch flow control with upstream back‑pressure and recovery (Blanc 6,660,300; Sindhu 6,493,347; plus the pre‑2001 switch‑ASIC art already on the record) — with the motivation supplied by the unavoidable per‑source queue imbalance created by any multi‑plane serial backplane and a finite set of known remedies.
• Claim 2 is the intermediate claim and is probably the easiest to invalidate on this record (per‑source queueing is dictated by striping; watermark overflow/underflow detection is the textbook technique).
• Claim 4 ("throttling back") and claim 11 (notify other components) are near‑verbatim flow‑control/notification steps.
• Claim 5 (K2) is the least vulnerable only in the sense that a specific reference may need to be paired for the 8b/10b ordered‑set practice; the claim is nonetheless a design choice over a finite set and is further weakened because the specification itself treats K0/K1/K2 as the standard reserved K‑codes of the serial standard it uses.
• The whole set is exposed by claim 1's breadth: it claims no striping, no cell format, no redundancy, no thresholds and no FIFOs.

What I could not verify, and what to pull next. (a) The verbatim claim text from the grant PDF/Patent Center (my source was a single mirror); (b) the filing dates of 6,842,422, 6,721,313, 6,697,359, 6,798,740, 6,654,370 and 6,660,300, which is dispositive of whether each is even available for § 103 at a Nov. 16, 2001 critical date; (c) the actual full disclosures of those references — I have titles/dates only, so every element mapping above is a candidate reading requiring a claim chart; (d) the prosecution history of 12/400,594 and its sibling 12/400,645 (US 7,948,872), which will show exactly what the examiner treated as the art of record and which limitation drove allowance — the fastest route to a defensible invalidity theory here (the family shows repeated Non‑Final/Final/Advisory actions across 09/988,066 and the 2009 continuations in the '580's own Other‑References list).