US 7062570

I'll search for authoritative records on this patent, including any 2026 litigation activity.

Let me pull the claim set and confirm any litigation/appeal records.

Let me retrieve the full claim set, including the earlier independent claims.

USPTO / CAFC Search Note

I searched for the specific number 7062570 (US 7,062,570) and did not find any 2026 Federal Circuit docket, appeal, or IPR tying to this patent number. ⚠️ Literal-ID caution: several 2026 results that string-match "7062570" are not this patent — e.g., JP 7062570 B2 (a Japanese patent for a light-transmitting panel frame, Nippon Steel Kenzai / Shutoko, registered 2022) and CAFC case numbers such as 2025-1470. Per your instruction, I have not auto-corrected or conflated these. The only litigation records surfacing for the current assignee are unrelated patents (e.g., T5.2 Ltd v. Citrix, S.D. Fla. 0:24-cv-62093; AttestWave v. Citrix, D. Del. 1:25-cv-00250, U.S. 7,895,643; K Mizra v. Citrix, S.D. Fla. 0:25-cv-60803, U.S. 8,234,705) — none assert 7,062,570.

Bibliographic Data (US 7,062,570 B2)

Sources: https://patents.google.com/patent/US7062570/en ; https://patents.justia.com/patent/7062570 ; https://www.freepatentsonline.com/7062570.html

Abstract (verbatim)

"The present invention is directed to a network switch that determines when specific content is hot and directs flow to one or more cache servers. The architecture can include a tag generator to generate unique tags corresponding to a server in a plurality of servers, a content pre-fetching algorithm to retrieve information before the information is requested by determining the hotness of the information, and a cache server in which stored information is configured based upon the relative degrees of hotness of the stored information."

Plain-Language Overview

The patent describes a "content director" — a Layer 2/3/4 Web switch placed in front of a server farm (origin servers, dynamic-content servers, cache servers). It assumes the farm's global IP address, intercepts client transaction requests, and decides where each request goes. It learns traffic patterns by snooping URLs, counts hits per URL/object, applies a "hot URL threshold," and once content is "hot," redirects flows for that content to a cache-server tier. It has two operating modes: a tagging mode (append a small server-identifying tag onto a server-generated cookie in the response, so later requests can be routed straight to the same server with minimal parsing) and a digesting mode (hash URL/cookie invariants into a digest that indexes a hot-URL table directly, avoiding linear table searches). Switch-cluster members share digest/tag information for redundancy and scale.

Independent Claims — What I Can Verify

Verified verbatim (Claim 43 — method): A method for switching transaction requests comprising: receiving a transaction response from a first source; parsing the response to locate a first field; determining a first tag identifying the first source; appending the tag to that field in the response; reassembling the response; forwarding it to the destination identified in the response; receiving a second transaction request; parsing that request for the first field; determining a digest value based on field information in that field; and storing selected information corresponding to the request at an address based on the digest value. In plain terms: stamp the outbound response with a server tag, and use a hash/digest of the request's URL-or-cookie field to write the request's state directly to a computed memory location (rather than searching a table).

Representative dependents: 44 (HTTP, hashing of URL/cookie), 45 (decrypt cipher-text request first), 46 (increment/decrement hit counter, compare to threshold, update timestamp), 47 (reverse-DNS to find network addresses for hot content), 48 (direct to cache server), 49 (existing connection → origin server; otherwise hot → cache), 50 (if cache can't serve, fall back to origin), 51 (transfer content from origin to cache).

Verified verbatim (Claim 52 — system, means-plus-function): A system for switching transaction requests among a plurality of servers, comprising: an input port; means for parsing the response to locate a first field; means for determining a tag identifying the first server; means for appending the tag to that field; means for reassembling; means for forwarding to the client; the input port receiving a second request; means for parsing the second request for that field; means for determining a digest value from that field; and means for storing information at an address based on the digest value. Dependent claim 54 adds a means for decrypting cipher-text requests.

So claims 43–54 are a tagging + digesting pair (method and apparatus), with the tag written on the response side and the digest used on the request side.

⚠️ Uncertainty on claims 1–42. Justia's published claim text was retrievable only from claim 43 onward. I could not retrieve the verbatim text of claim 1 or any independent claims numbered below 43 with high confidence, so I will not paraphrase them as if verbatim. Based on the abstract and the specification's stated aspects, claim 1 and/or other early independent claims most likely recite (a) a network flow switch/content director with a tag generator producing unique per-server tags, (b) content pre-fetching driven by relative hotness of sub-URLs, and/or (c) a cache server whose stored content is positioned/configured according to degrees of hotness — but this is inference, not verified claim language. Total claim count is also not confirmed from the sources retrieved (at least 54 claims exist).

Related Notes

• The tag is described as much smaller in bit size than the cookie, and is only appended when an outbound response passes through the switch; cookie path-restriction rules (domain/path/max-age, most-restrictive path) govern server persistence.
• Digests are computed with a hash h(K) → L mapped into a table of size M, with circular-linked-list collision chaining and a cache identifier (shown as an MD5 value) added so a cluster peer can locate which local cache holds an object.
• The specification cites Cache Array Routing Protocol (CARP) for distributing cacheable URL objects among cache proxies.
• The patent's asserted advantage over Port-80 redirecting switches is wire-speed switching for cache hits and most cache misses, with hot-URL/hot-IP identification to raise cache hit rates.

If you need the verbatim claims 1–42, the authoritative sources are the USPTO Patent Center/Patent Public Search record for 09/921,460 or the granted PDF at https://patents.google.com/patent/[US7062570B2](/patent/US7062570B2)/en — I was unable to pull those specific claim texts in this session, so I am flagging that as an open gap rather than filling it in.

I'll research litigation involving US patent 7062570. Let me run several targeted searches.

Let me search more targeted sources.

Summary Finding

I could not identify any litigation that asserts or otherwise involves U.S. Patent No. 7,062,570. Based on the searches I was able to complete, the record shows this patent being cited as prior art in other patents' prosecution and appearing in the assignee's portfolio listings, but I found no complaint, ITC investigation, IPR/PGR, or appeal naming 7,062,570 as a patent-in-suit.

Because I was working under a tool-call limit and my search of the Unified Patents litigation portal for this specific number did not complete, I want to be explicit: this is a "no evidence found" conclusion, not a certifiable "never litigated" conclusion. A definitive negative would require a full PACER/RECAP docket sweep plus the Unified Patents and Docket Navigator litigation databases.

What the patent is (for identification)

The expiration date is a relevant negative signal: with the patent expired and no damages window remaining for prospective infringement, it has little value as a suit vehicle today, which is consistent with the absence of enforcement activity I found.

Cases I checked and excluded (different patents — NOT this patent)

These are the closest-looking hits; each involves an Avaya- or Citrix-related party but asserts different patent numbers, so they must not be reported as litigation over 7,062,570:

1. SolutionInc Limited v. Avaya Inc. — D. Del., No. 1:20-cv-00185-LPS, filed 2020-02-06. Patent asserted: 7,526,538 (per the Rule 3 report to the Commissioner). Not 7,062,570.
2. AttestWave LLC v. Citrix Systems, Inc. — D. Del., No. 1:25-cv-00250-RGA, filed 2025-03-05. Patent asserted: 7,895,643. Not 7,062,570.
3. T5.2 LTD v. Citrix Systems, Inc. — S.D. Fla., No. 0:24-cv-62093-JB, filed 2024-11-05. Patents asserted relate to centralized graphical application servers (the '147, '192, '568, '922, '146 family). Not 7,062,570.
4. Lexos Media IP, LLC v. Amazon.com, Inc. et al. — E.D. Tex., No. 2:22-cv-00169-JRG (and member cases). Unrelated patents. Not 7,062,570.

Where 7,062,570 actually appears

• Assignee/portfolio listings: Plainsite's profile for "Avaya Technology Corporation" lists the patent among its holdings (i.e., an assignment/portfolio record, not a docket).
• Prior-art citations: Justia and Google Patents show 7,062,570 cited as a reference in other patents (e.g., U.S. 7,697,536; 7,881,332; 7,987,323) — ordinary prosecution citations, not litigation.
• Prosecution-family relationship: The specification states it is related to two concurrently filed applications, "Intelligent Demand Driven Recognition of URL Objects in Connection Oriented Transactions" and "Non-Intrusive Multi-plex Transaction Persistency and Secure Commerce Environments." These are sibling applications, not litigation.

Important caveats

• Searches that were inconclusive: My query to the Unified Patents litigation portal for this specific number returned a generic portal page for a different patent (US 5,923,885 A), so that source was not usefully searched. I was unable to run the additional docket-level checks I would want.
• A "no hits" result cannot rule out: (a) cases where 7,062,570 was one of many asserted patents in a large NPE portfolio suit whose caption doesn't name it; (b) sealed or dismissed-without-prejudice actions with sparse public dockets; (c) older Avaya-era declaratory judgment actions.
• Recommended verification steps: Run 7,062,570 through Docket Navigator or Lex Machina "patent-in-suit" search, query the USPTO PTAB API for IPR/PGR petitions naming it, and check RECAP/CourtListener for dockets where it appears as an exhibit or asserted patent.

Answer to your question, restated

No known litigation involving US Patent 7,062,570 was found. I recommend confirming with a dedicated patent-litigation database before relying on this as a final negative, particularly for any multi-patent portfolio suits.

I'll verify the structured "no proceedings" finding against public sources before reporting.

Let me run additional targeted checks for any AIA trial naming this patent number.

One search hit referenced a PTAB petition discussing a "'570 patent" with cache-server claims. Let me identify that patent precisely — it may or may not be ours.

Proceedings overview

Total AIA trial proceedings on file for US 7,062,570: 0 — zero active, zero in which claims were invalidated, zero in which claims were sustained, zero settled, zero institution denials. The bottom line for a defendant is uncomfortable rather than reassuring: this patent has never been tested at the PTAB at all, so there is no FWD to point to, no canceled claim to exploit, and no IPR-based estoppel or invalidity finding to inherit — but it is also expired (adjusted expiration 2023-09-07), so the practical exposure is limited to a § 286 six-year back-damages window, and the absence of any petition in 20+ years of enforceability is itself strong evidence that nobody has ever found this patent worth attacking.

I reached this conclusion from two independent sources: the canonical USPTO Open Data Portal structured block in the prompt (which reports no AIA proceedings on file), and web searches of PTACTS/PTAB materials, which surfaced no petition, institution decision, FWD, or appeal naming 7,062,570 as the challenged patent. Searches that string-matched "7062570" returned only the unrelated items in the disambiguation table below.

Because there are no proceedings, the per-proceeding ### template has no entries to populate. I will not manufacture any.

Disambiguation: every "'570" / "7062570" hit I checked, and why none is this patent

A defendant's paralegal will find all of these and they must not be carried into a brief as IPRs on US 7,062,570.

That last row is the only item in the entire search that came close enough to matter, and it fails on the merits of its own quoted claim language.

Strategic summary

Claim status: UNTESTED ACROSS THE BOARD. No claim of US 7,062,570 has been canceled, confirmed, or even construed by the PTAB. The full issued claim set — including the method claim 43 and the means-plus-function system claim 52 identified earlier in this analysis, and claims 1–42 that remain unverified — stands exactly as granted on 2006-06-13, subject only to whatever (if anything) happened in ex parte reexamination, which I did not exhaustively search and therefore will not characterize. There is no FWD to quote and no disposition to link.

Estoppel landscape: empty, and that cuts in a defendant's favor. Because no third party ever filed a petition, no petitioner is subject to the 35 U.S.C. § 315(e)(2) bar. A defendant today is therefore free to raise any § 102/§ 103 ground, and — unlike the typical NPE scenario — is not collaterally limited by someone else's narrower prior-art choices. The corollary is that there is also no earlier institution decision, expert declaration, or Board claim construction to borrow. Everything would have to be built from scratch, against a 2000-priority patent whose specification cites (per FreePatentsOnline's reference list) Alteon/Apache-style content switching, Apostolopoulos Design, Implementation and Performance of a Content-Based Switch (INFOCOM 2000), and Ross Hash-Routing for Collections of Shared Web Caches (IEEE Network, Nov./Dec. 1997) — i.e., the patent sits in a fairly crowded 1997–2000 content-switching art space, which is where an invalidity case would have to be made.

Pattern signals: none. No repeat petitioner, no Unified Patents or other defensive aggregator involvement, no parallel PTAB appeal, no IPR↔district-court stay dynamic. That is the expected signature of a patent that was never asserted: I found no litigation in this analysis either, and the patent expired 2023-09-07. AIA trial statistics are driven by assertion; a patent that is never asserted almost never attracts an IPR, which is the most parsimonious explanation for the null result here — as opposed to any inference that the claims are unusually strong.

What is legally unavailable going forward. PGR was categorically unavailable from the start: it requires an effective filing date on or after 2016-09-16 (AIA) / 2013-03-16 (transition), whereas this patent claims priority to provisional 60/223,087 filed 2000-08-04. CBM review under AIA § 18 was limited to "covered business method" patents (financial-services claims) and its transitional program sunset on 2020-09-16 — a Layer 2/3/4 Web-switch patent would not have qualified in any event. IPR remains the only theoretical vehicle, but with the patent expired, a petitioner would be spending six figures to invalidate claims that can support at most back damages, which is why the zero-count is unlikely to change.

Recommended next steps

1. Do not represent that this patent has been IPR-tested — or IPR-cleared. If you are defending an assertion, the honest framing is: claims as issued, never construed, expired 2023-09-07, no PTAB or district court record.
2. Verify the ex parte reexamination gap. I searched PTAB/AIA materials only. Before relying on "claims stand as issued," run the USPTO Patent Public Search / PatentCenter record for application 09/921,460 for any 90/xx,xxx reexamination certificate. This is the single largest open item in my analysis.
3. Confirm the ambiguous "'570" petition. If you need certainty on the PTACTS document at https://ptacts.uspto.gov/ptacts/public-informations/petitions/1459538, pull the front page of the petition to read the challenged patent number verbatim. My reading of the quoted limitations is Parallel Networks' US 7,571,217, but I am flagging it as unconfirmed rather than asserting it.
4. Assertion-defense posture. If a demand letter arrives citing 7,062,570: (a) check the date of the accused conduct against the 2023-09-07 expiration and the § 286 six-year lookback; (b) note there is no litigation or PTAB record to satisfy any "objectively reasonable" invalidity-defenses showing, infra; and (c) treat claim 1's language as unverified — earlier in this analysis I could retrieve verbatim text only from claim 43 onward, so obtain the granted claims from the USPTO PatentCenter record or https://patents.google.com/patent/[US7062570B2](/patent/US7062570B2)/en before responding on the merits.
5. Sibling-patent check. If the same family is being asserted, the co-filed sibling "Intelligent Demand Driven Recognition of URL Objects in Connection Oriented Transactions" (issued as US 7,228,350, Hong/Chu/Jaswa, filed 2001-08-03) should be run through the PTAB API separately — family members sometimes carry the IPR activity that the asserted patent lacks.

Caveats. This is a "no evidence found" conclusion from the ODP structured data plus web searching, not a certified negative. It cannot rule out (a) a petition in which 7,062,570 was one of many challenged patents in a multi-patent filing whose caption does not name it, (b) sua sponte Director review or sua sponte PTAB actions not surfaced by search, or (c) pre-AIA interferences and ex parte reexaminations, which are outside the AIA-trial scope requested here. The recommended confirmation path is the USPTO PTAB API/PTACTS search for the patent number, plus Docket Navigator or Lex Machina "patent-in-suit" and "PTAB" searches.

I'll reconstruct the assignment chain, starting with targeted searches for reel/frame records and correspondents.

Let me pin down reel/frame numbers for the key transfers and check the current prosecution correspondent.

Assignment Chain & NPE Analysis — US 7,062,570

Sourcing note before the analysis. I was able to reconstruct the chain from the Google Patents legal-events register (which mirrors the USPTO Assignment Center records), but I could only independently verify three reel/frame numbers (012759/0141, 044891/0439, 062113/0001). The remaining recordings are documented by date, conveyance type, assignor, and assignee but their reel/frame stamps did not surface in this session. I flag every such gap below rather than filling it in. Authoritative verification: USPTO Assignment Center and the Google Patents legal-events tab for application 09/921,460.

Inventors

Pattern check — inventorship continuity: All three inventors are the same-named inventors on the two concurrently filed sibling applications ("Intelligent Demand Driven Demand Recognition of URL Objects in Connection Oriented Transactions" and "Non-Intrusive Multi-plex Transaction Persistency and Secure Commerce Environments," per the specification's CROSS REFERENCE section). That is a coherent in-house R&D team filing a three-application cluster on one filing date — not a classic red flag.

Pattern check — inventor departure ⇒ fire-sale: Not determinable. I found no evidence of the inventors leaving Avaya within 12 months of the 2001-08-03 filing, and I did not fabricate one. Note the assignment to Avaya Technology Corp. was recorded roughly 14 months after filing, which is unremarkable for a company taking confirmatory rights in employee inventions.

Original assignee

Avaya Technology LLC (originally Avaya Technology Corp.), the patent-holding subsidiary of Avaya Inc.

• Primary line of business: Avaya is an enterprise communications company (IP telephony, contact center, unified communications), spun out of Lucent Technologies' enterprise networking business in 2000 — which is why this server-farm/web-switch patent sits alongside telephony patents in the same portfolio.
• Status of the original assignee: Chapter 11 twice — first in January 2017, again in February 2023 (reorganized/emerged ~2023 under new ownership). The 2017 case is directly visible in this patent's own record via the 2017-12-15 "BANKRUPTCY COURT ORDER RELEASING ALL LIENS" event.
• Did Avaya ship a product embodying the claims? Unclear / not confirmed. Avaya's commercial line was enterprise communications, not a standalone content-director/L7 web switch. I found no evidence Avaya commercialized this specific architecture, and it had already divested the patent to Citrix in 2009 — eight years before its first bankruptcy. I am flagging this as unresolved rather than asserting a product existed.

Current owner: Citrix Systems, Inc., now part of Cloud Software Group (the Vista Equity Partners / Evergreen Coast Capital entity formed after the 2022 Citrix take-private and merger with Tibco Software). Citrix is a genuine operating company, and its NetScaler / Citrix ADC product line is a Layer‑7 application delivery controller doing precisely the load-balancing, content switching, SSL offload and caching that these claims describe — so a product embodying the claim space plausibly exists. I have not, however, verified that NetScaler reads on these specific claims.

Assignment timeline

Records below are from the Google Patents legal-events register for application 09/921,460 (mirror of USPTO Assignment Center). Reel/frame shown only where independently verified.

2002-04-05 (executed) / recorded 2002-04-09 — Reel 012759/0141 ✅ verified

• Conveyance: Security Agreement (grant of security interest)
• Assignor: Avaya Technology Corp.
• Assignee: The Bank of New York (collateral agent)
• Correspondent: not surfaced in this session.
• Context: Securitization — lien on the portfolio; this is the only pre-2009 encumbrance expressly tied to this patent's own record (later released, see 2017 entry).

2002-10-15 (recorded) — reel/frame not captured

• Conveyance: Assignment of inventors' interest
• Assignors: Jaswa, Vijay; Hong, Jack; Chu, Albert Bonyao
• Assignee: Avaya Technology Corp.
• Correspondent: not captured.
• Context: Internal/confirmatory — standard employee-inventor assignment to the operating parent.

2007-11-27 (recorded) — reel/frame not captured

• Conveyance: Security Agreement
• Assignors: Avaya Technology LLC; Avaya, Inc.; Octel Communications LLC; VPNet Technologies, Inc.
• Assignee: Citibank, N.A. (administrative agent)
• Context: Securitization — cross-portfolio lien securing Avaya's 2007 credit facility.

2007-11-28 (recorded) — reel/frame not captured

• Conveyance: Security Agreement
• Assignors: same four Avaya entities
• Assignee: Citicorp USA, Inc. (administrative agent)
• Context: Securitization — parallel first-lien facility.

2008-06-27 (recorded) — reel/frame not captured

• Conveyance: Reassignment
• Assignor: Avaya Technology LLC
• Assignee: Avaya Inc
• Context: Internal reorganization — movement of portfolio among Avaya affiliates.

2008-12-29 (recorded) — reel/frame not captured

• Conveyance: Conversion from Corp to LLC
• Assignor: Avaya Technology Corp.
• Assignee: Avaya Technology LLC
• Context: Change of name/entity form only — no change in beneficial ownership.

2009-10-26 (recorded) — four releases, reel/frames not captured

• Conveyance: Release by Secured Party (×4)
• Assignors/releasing parties: Citicorp USA, Inc.; Citibank, N.A.; Silicon Valley Bank; The Bank of New York
• Assignee/releasee: Avaya Inc. (and Avaya Technology Corp. / Avaya)
• Context: Lien clearance — the entire security stack is discharged in a single day, immediately before the sale below. This is the classic "make clean title" step.

2009-10-27 (recorded) — reel/frame not captured

• Conveyance: Assignment of interest
• Assignor: Avaya Inc.
• Assignee: Citrix Systems, Inc.
• Correspondent: not captured.
• Context: Divestiture / portfolio acquisition by an operating company — the day after the liens cleared. This is the ownership-transfer event that matters; it is not a transfer to a shell.

2017-12-15 (recorded) — Reel 044891/0439 ✅ verified

• Conveyance: Bankruptcy Court Order releasing all liens (specifically the security interest recorded at Reel/Frame 012759/0141)
• Assignor (releasing party): The Bank of New York
• Assignee: Avaya Inc. (formerly known as Avaya Technology Corp.)
• Context: Bankruptcy — but not a fire-sale of this patent. Avaya's 2017 Chapter 11 released a 2002 legacy lien. The patent itself had been sold to Citrix eight years earlier, so it was never part of the Avaya estate.

2018-01-09 (recorded) — reel/frame not captured

• Conveyance: Release by Secured Party
• Assignor (releasing party): Citicorp USA, Inc.
• Assignees: Octel Communications LLC; Avaya, Inc.; Sierra Holdings Corp.; VPNet Technologies, Inc.; Avaya Technology, LLC
• Context: Bankruptcy exit-financing cleanup of the 2007 lien stack.

2022-09-30 (recorded) — reel/frame not captured

• Conveyance: Security Interest
• Assignor: Citrix Systems, Inc.
• Assignee: Wilmington Trust, National Association
• Context: Securitization tied to the Citrix take-private (Vista/Evergreen) — this date is the closing of the going-private transaction.

2022-10-07 (recorded) — three recordings, reel/frames not captured

• Conveyance: Patent Security Agreement (Bank of America, N.A., collateral agent) / Second Lien Patent Security Agreement (Goldman Sachs Bank USA, collateral agent) / Patent Security Agreement (Wilmington Trust, N.A., notes collateral agent)
• Assignors: Citrix Systems, Inc.; Tibco Software Inc.
• Context: Securitization — the debt package financing the Citrix/Tibco combination into Cloud Software Group.

2023-04-14 (recorded) — Reel/Frame 062113/0001 referenced ✅ cited in the record

• Conveyance: Release and Reassignment of Security Interest in Patent (reel/frame 062113/0001)
• Assignor (releasing party): Goldman Sachs Bank USA, collateral agent
• Assignees: Cloud Software Group, Inc. (f/k/a Tibco Software Inc.); Citrix Systems, Inc.
• Context: Refinancing — second-lien release following the Cloud Software Group reorganization. Note: the reel/frame cited (062113/0001) is the released Goldman instrument; I did not independently confirm the reel/frame of the releasing record itself.

2023-04-14 (recorded) — reel/frame not captured

• Conveyance: Patent Security Agreement (new notes collateral grant)
• Assignors: Citrix Systems, Inc.; Cloud Software Group, Inc. (f/k/a Tibco Software Inc.)
• Assignee: Wilmington Trust, National Association (notes collateral agent)
• Context: Securitization — replacement notes lien.

2023-09-07 — Adjusted expiration (term ends; patent expired)

Correspondent of record (prosecution): FOLEY & LARDNER LLP / CITRIX SYSTEMS INC., Washington, DC — this is the current attorney/agent of record on the granted patent, updated to the present owner. It is a prosecution correspondent, not an assignment-recording correspondent, and I found no recurring assignment correspondent across the links. Signal 3 therefore cannot be substantiated.

Timeline diagram

timeline
    title Ownership of US 7062570
    2000 : Provisional filed by Avaya
    2001 : Application filed by Avaya Technology LLC
    2002 : Inventors assign to Avaya Technology Corp
         : Bank of New York security interest
    2006 : Patent issued
    2007 : Citibank and Citicorp security agreements
    2008 : Conversion to Avaya Inc
    2009 : Liens released then sold to Citrix Systems
    2017 : Avaya bankruptcy releases legacy lien
    2022 : Citrix take private security interests
    2023 : Cloud Software Group reorg
         : Patent term ends

NPE / troll-pattern signals

Verdict

Operating-company assertion — with the express caveat that no assertion activity was identified.

Justification: the entire chain is a clean operating-company lineage — Avaya Inc. assigned to Citrix Systems, Inc. on 2009-10-27, immediately after all four liens were released on 2009-10-26 (Citicorp, Citibank, Silicon Valley Bank, Bank of New York). From that point forward every recorded assignment is either a bank as collateral agent (Wilmington Trust, Bank of America, Goldman Sachs) or the post-2022 Cloud Software Group reorganization, none of which resembles a shell-entity or NPE transfer. The only bankruptcy-adjacent event — the 2017-12-15 order releasing the lien recorded at Reel 012759/0141, recorded at Reel 044891/0439 — concerned a creditor's lien on an asset Avaya had already divested, so it does not convert this into a fire-sale chain.

⚠️ Caveat on the verdict label: the "assertion" prong is unmet — I found no infringement suit naming 7,062,570 (consistent with the earlier litigation summary), and the patent expired 2023-09-07. Read this verdict as "operating-company-owned, not an NPE, no enforcement activity found."

Verify at: USPTO Assignment Center — search patent 7,062,570 · Google Patents legal events · USPTO Patent Public Search, app 09/921,460

Open gaps I could not close in this session (stated plainly, not filled): reel/frame numbers for the 2002-10-15 inventor assignment, the 2007-11-27/28 Citibank and Citicorp security agreements, the 2008 reassignment and 2008 name conversion, the four 2009-10-26 releases, the 2009-10-27 Avaya→Citrix assignment, the 2018-01-09 release, and the three 2022-10-07 / two 2023-04-14 security agreements. Pulling those stamps — and, in particular, the recording correspondent on the 2009 Avaya→Citrix assignment — is the single highest-value next step if you need to stress-test the "no repeat correspondent" finding.

I'll research the specific prior-art citations for US 7,062,570 from authoritative patent databases.

I have a partial list from Justia. Let me retrieve the complete front-page references from the patent PDF and cross-check with Google Patents.

Let me get the complete citation list from the patent's front page and Google Patents' citation section.

Prior Art Analysis — US 7,062,570 B2

0. Patent identified (record confirmation)

The number 7062570 resolves in the patent databases to US 7,062,570 B2, "High performance server farm with tagging and pipelining," App. No. 09/921,460, filed 2001-08-03, priority to provisional 60/223,087 filed 2000-08-04, granted 2006-06-13. Inventors Jack Hong, Albert Bonyao Chu, Vijay Jaswa. This is the same record already summarized in the earlier sections, so I won't restate the bibliographic table.

⚠️ A caution already flagged in the prior sections and still true here: the literal string "7062570" also resolves to JP 7062570 B2 (Nippon Steel Kenzai / Shutoko Kōsoku Dōro, a light-transmitting panel frame, registered 2022-05-06, application 特願2018-182068, publication 特開2020-51137). That is a different, unrelated patent and is not the subject of this analysis. Nothing below comes from it.

⚠️ Tool-budget disclosure: this session hit its tool-call ceiling before I could pull the granted-PDF front page directly or verify each reference's filing date. The citation set below was assembled from Justia's "Referenced Cited" rendering of the 7,062,570 front page and cross-checked against FreePatentsOnline's rendering of the same "References Cited" block. Filing dates are not verified for most entries; where a §102(e) determination depends on the reference's U.S. filing date, I say so rather than guessing.

1. The critical dates that drive every §102 call

App. 09/921,460 is pre-AIA (filed 2001-08-03), so pre-AIA 35 U.S.C. §102(a)–(g) governs. Three dates matter:

Consequence for grouping: anything issued/published before 1999-08-04 is a clean §102(b) candidate; anything issued/published after 2000-08-04 can only be prior art under §102(e) (or §102(g)), which depends on its own U.S. filing date; the window in between turns on which of the two bar dates applies. I have applied this grouping below.

2. Complete "References Cited" set for US 7,062,570

2(a) U.S. Patent Documents

⚠️ List-integrity flag: the Justia rendering of this block begins at 5,710,883. I could not positively confirm whether additional earlier U.S. references (5,xxx or lower) sit above it on the granted front page; the granted PDF front page is the check.

2(b) Foreign patent documents

⚠️ I report these exactly as retrieved. The "PCT/US 01/24313" entry as rendered is ambiguous (an application number in a "WO" slot), and I did not resolve it. Given the Aug-3-2001 co-filing family, this is very likely the applicant's own sibling filing being listed in the IDS, not third-party prior art.

2(c) Non-patent literature (NPL) — all printed publications

3. Claim-by-claim §102 relevance

Critical limitation on this section: as noted in the earlier summary, only claims 43–54 could be verified verbatim. Claims 1–42 remain unretrieved, and the verified content of claim 43 (tag-on-response + digest-addressed storage) and claim 52 (means-plus-function counterpart) is what I can map against. Anticipation requires a single reference to disclose every limitation, so what follows is a plausibility ranking, not a made-out §102 rejection.

3(a) Claims 43 / 52 and their dependents — the "tag + digest" combination

The independent claims require, together: (i) parse an outbound response for a field; (ii) determine a tag identifying the source server; (iii) append the tag to that field; (iv) reassemble and forward; (v) receive a subsequent request; (vi) parse that request for the field; (vii) determine a digest value from the field; and (viii) store request-corresponding information at a digest-derived address.

• Ross, "Hash-Routing for Collections of Shared Web Caches" (Nov/Dec 1997) — the single most on-point NPL item for limitation (vii)/(viii). Hash-routing maps a URL hash to a cache, i.e., computing an address from request content rather than searching a table. Best candidate to anticipate a claim drawn narrowly to hash-derived addressing or to the CARP-based cache selection described in the spec. It does not appear to disclose (i)–(iv), the server-identity tag.
• RFC 2109 (Feb 1997) — supplies the cookie/Set-Cookie field and domain/path/max-age mechanics that claim 43's "first field" presupposes. It is background to (i) and (iii), but does not disclose a switch-generated server-identity tag. A §102 rejection resting on RFC 2109 alone would fail on (ii)–(iii); it is more realistically §103 fodder.
• RFC 2105 — Cisco Tag Switching (Feb 1997) — cited because of the word "tag," but tag switching is an MPLS-style forwarding-label scheme, not server-identity tagging of cookies. Superficial terminology overlap only; I would not expect it to anticipate claims 43/52.
• RFC 2334 — Server Cache Synchronization Protocol (Apr 1998) — bears on the specification's cluster-wide sharing of digest/tag state (the "cache digests from its peers" feature), which underpins dependent claims and the redundancy/scalability description. Not an anticipation reference for 43/52.
• Franklin et al. (1997) and Yin et al. (1999) — cache-consistency/revalidation art. Relevant to the expiration-timer/revalidation aspects, i.e., more the dependent-claim and §103 territory than the independent claims.
• 6,108,703 Leighton et al. (issued 2000-08-22; §102(e) only) — demand-based content delivery/replication. Potentially relevant to claim 43's digest step only if it computes a content-derived address; I did not verify its disclosure this session, so I flag this as an unconfirmed possibility rather than a mapping.
• 6,134,583 Herriot (2000-10-17; §102(e)) — HTTP request/proxy handling; candidate against the parsing steps (i),(vi) but not the tag or digest steps.
• 2002/0112154 Wallace, Jr. and 6,772,333 Brendel — surfaced in the same citation block; given their dates (2002/2004) they are §102(e)-only and cannot be statutory bars. Subject matter unverified.

3(b) Claims 1–42 (inferred scope — see earlier ⚠️)

The earlier summary flagged that these are most likely directed to (a) the tag generator producing unique per-server tags, (b) content pre-fetching driven by relative sub-URL hotness, and (c) a cache server whose stored content is arranged by degree of hotness. On that inferred scope:

• Alteon "Virtual Matrix Architecture" (Apr 2000) and the two ArrowPoint/Cheng Wu "What is Web Switching?" papers (Oct 1999) — closest cited art on web-switch content-based load balancing and cache tier insertion. These are the references I would expect the examiner to have used for any pre-fetching or hot-content-routing limitation.
• Apostolopoulos et al., INFOCOM 2000 — "Content-Based Switch" is directly on the parse-URL-and-route architecture.
• 6,108,703 Leighton et al. — the cited U.S. patent most likely to bear on hotness/demand-triggered replication into a cache tier.
• Cohen et al. (Aug 1998) — server-assisted cache replacement based on popularity; the best cited-art candidate against a claim reciting content placement/positioning by degree of hotness.
• Arlitt (SIGMETRICS 1996) — workload-characterization/invariants; bears on the "tunable hot URL threshold" concept.
• Liebrmann (Jan 2000) — general load-balancing survey; background.

Because claims 1–42 are unverified, I am not asserting any specific §102 mapping to them. Doing so would be inference dressed as finding.

4. What this list does and does not prove

1. These are "References Cited," not the examiner's rejection grounds. The block mixes applicant-submitted IDS art (the ArrowPoint/Alteon/Apostolopoulos/Network Magazine items are classic competitor-product IDS submissions) with examiner-cited art. The printed front page does not tell you which reference was applied under §102 versus §103, or against which claim. That information lives in the file wrapper (Patent Center for App. 09/921,460), which I could not open this session.
2. §102(a)/(b) versus §102(e) is decisive for roughly two-thirds of the list. Fifteen of the nineteen U.S. entries issued after the 2000-08-04 priority date and therefore cannot be §102(b) bars; they can only reach the claims via §102(e) on their own earlier U.S. filing dates. I did not verify those filing dates, so I am not asserting that any of 6,108,703; 6,134,583; 6,304,973; 6,330,606; 6,438,652; 6,718,387; 6,763,370; 6,772,333; 6,799,202; or 6,862,624 is §102(e) art — only that §102(e) is the only route available to them.
3. One entry is not prior art. The "USPTO Office Action dated Jul 15, 2005 in U.S. Appl. No. 09/921,832" is a prosecution document from a co-filed sibling application (the Aug-3-2001 family; the sibling that the earlier sections identify as issuing as US 7,228,350). It is an IDS item, not §102 art.
4. "Cited By" ≠ prior art. Separately from the above, 7,062,570 appears as a forward citation in later patents (e.g., US 7,697,536; 7,881,332; 7,987,323; US 8,145,768; US 7,706,409). Those are references that cited this patent; they are not prior art against it and are outside the scope of your question.
5. Nothing here is litigation. There is no §102 determination to be had from a docket, and as the earlier litigation section concluded, no case asserting 7,062,570 was located.

5. Recommended verification steps (open items I could not close)

Bottom line: the prior art actually cited on the face of US 7,062,570 is a 19-entry U.S. patent block (5,710,883 through 2004/0230566), two foreign/WO entries, and sixteen printed publications (RFC 2109, RFC 2105, RFC 2334, five academic papers, three ArrowPoint web-switch/product documents, one Alteon document, one Network Magazine article, one INFOCOM 2000 paper, and one sibling-application Office Action). Of these, the §102(b)-eligible core — the five 1998–mid-1999 U.S. patents (5,710,883; 5,737,536; 5,764,920; 5,825,772; 5,923,885) and the sixteen NPL items — is where a genuine anticipation case would have to be built, with Ross (1997) on digest/hash-derived addressing, Apostolopoulos (2000) and the ArrowPoint/Alteon materials on content-based web switching, Cohen (1998) on popularity-based cache placement, and Leighton 6,108,703 on demand-triggered replication as the four most consequential. I could not verify full reference texts or filing dates, and claims 1–42 remain unretrieved, so the claim mapping above is expressly a ranked hypothesis, not a made-out §102 rejection.

I'll ground this analysis in the actual prior-art references cited on the patent's record. Let me pull them.

Let me pull the actual citation lists from the patent record and the key art-space references.

Let me try once more to get the patent's own citation table, since that is what a § 103 analysis should anchor to.

§ 103 Obviousness Analysis — US 7,062,570 B2 ("High performance server farm with tagging and pipelining")

0. Scope note — a required flag before anything else

Your instruction was to "use the results from the Prior Art section of this page." No Prior Art section was supplied in my inputs. The material I received consists of (a) the full patent text (claims not included in the fetched HTML), and (b) the previously generated summary / litigation / PTAB sections. The earlier sections referenced a FreePatentsOnline "reference list" (Alteon/Apache-style content switching, Apostolopoulos INFOCOM 2000, Ross IEEE Network 1997) but did not reproduce the patent's actual [56] References Cited block.

I attempted to retrieve the patent's verbatim citation table (Google Patents / Justia / FPO) and could not confirm the exact 1449 list in this session. So, per the operating rule against fabrication, I am labeling every reference below by how I can ground it:

⚠️ Inconsistency to resolve: the earlier-generated "PTAB challenges" section stated the specification's reference list includes Apostolopoulos and Ross. My searches confirmed both papers exist and predate the priority date, but I could not independently confirm they sit in the 7,062,570 record. Treat the specific 1449 membership as an open item.

1. Governing law and the person of ordinary skill

Statute. The application was filed 2001-08-03, claiming § 119(e) benefit of provisional 60/223,087 filed 2000-08-04. Both dates precede 2013-03-16, so pre-AIA 35 U.S.C. § 103(a) governs. Prior art must predate the effective filing date — 2000-08-04 for subject matter carried in the provisional, otherwise 2001-08-03.

POSITA (my proposed construction). A person with a bachelor's in EE/CS (or equivalent) and 2–4 years of experience designing or deploying L4–L7 content switches, HTTP load balancers, or web caching/reverse-proxy systems, familiar with TCP/IP, HTTP/1.0–1.1, cookie persistence, DNS, hash tables, and the 1997–2000 content-delivery literature.

Governing obviousness framework. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): a combination is obvious where (i) the elements were known, (ii) there was a known problem for which the combination was a known solution, (iii) the combination is a predictable variation of a known technique, or (iv) there was a design incentive / market pressure with a reasonable expectation of success.

Critical posture facts (from earlier analysis, carried forward).

• Patent expired 2023-09-07; no damages exposure past the § 286 six-year lookback.
• Zero litigation, zero AIA trials — so there is no claim construction, no FWD, no estoppel, and no prosecution-history narrowing to lean on. Any § 103 case must be built de novo.
• Earlier sections correctly identified claims 43 and 52 as the verified independent claims (tagging + digesting), with claims 1–42 unverified. I therefore analyze claim 43/52 in detail and treat 1–42 by feature family, flagged as such.

2. The contested subject matter, decomposed into limitations

Per the earlier verified claim text and the specification:

Claim 43 (method) — decompose into four functional clusters:

• A. Response-side interception + field parse (receive response; parse to locate a "first field," e.g., cookie).
• B. Tagging (determine a tag identifying the source server; append tag to that field; reassemble; forward to destination).
• C. Request-side parse (receive second request; parse for the same field).
• D. Digest addressing (determine a digest value from field info; store selected information at an address based on the digest value).

Claim 52 (system, § 112(f) means-plus-function) — the same four clusters expressed as means; corresponding structures = IFS 200, cache processor 208, digest generator 216, cache 212.

Claims 1–42 [UNVERIFIED] — the abstract + SUMMARY indicate three likely independent families:

• F1: tag generator producing unique per-server tags appended to URL/cookie;
• F2: content pre-fetching driven by relative hotness of sub-URLs referenced in returned content;
• F3: cache-server content positioning according to degree of hotness (incl. geographic/tiered placement).

Everything below maps prior art to A/B/C/D (claims 43/52) and to F1/F2/F3 by family.

3. The prior-art landscape

References that are NOT prior art — flag to avoid error:

• US 7,228,350 ("Intelligent Demand Driven Recognition of URL Objects in Connection Oriented Transactions," same inventors Hong/Chu/Jaswa, filed the same day) is a sibling in the same family / same inventive entity. Under pre-AIA § 102(e) the reference must be "by another"; a co-filed application by the identical inventive entity is not § 102(e) prior art against 7,062,570. Do not chart it as such. (It remains relevant only as corroboration of the contemporaneous state of the art.)

4. The elephant: the patent admits most of the inventive story

This is the single most important observation for the § 103 analysis, and it is not speculation — it is the patent's own Background.

[ADMITTED-2] states that prior-art Web switches already:

1. looked "into the HTTP payload down to the URL and cookie";
2. used "sticky connections";
3. "track[ed] content requests and … predict[ed] hot content before a server becomes overwhelmed";
4. "dynamically replicate[d] hot content to a Web cache and br[ought] the cache into the load balancing rotation"; and
5. "track[ed] which servers have delivered specific content and sen[t] new requests for that content directly to the appropriate server, resulting in improved server cache coherency."

That is, the problem statement and the broad solution of the challenged patent — detect hot content → route hot flows to cache servers → keep later requests pinned to the server that served the content — is conceded as known. The patent's asserted delta is narrower: the specific mechanism (a small tag appended to the cookie on the response side; a digest computed on the request side that directly addresses the state store) and the cache-organization/prefetch refinements.

This concession massively lowers the § 103 bar and frames every combination below.

5. Combination grounds

Combination 1 — Core § 103 ground for claim 43 (and its F1/system counterpart, claim 52)

Primary refs: ArrowPoint CSS cookie-persistence switch [FIELD-ART-1] (or Alteon WebOS [FIELD-ART-2]) in view of Cache Digests [FIELD-ART-7] and conventional hash-table direct addressing [FIELD-ART-10].

Motivation to combine ([FIELD-ART-1/2] + [FIELD-ART-7/10]):

• Same field, same problem. Both references address URL/cookie-driven content routing in web-switch/cache deployments; combination requires no field-switch by the POSITA.
• The patent's own admitted problem supplies the motivation. The specification concedes: "Conventional flow switches using a cache typically maintain a table of objects … The switch searches through the table from top to bottom … Substantial computational resources are consumed for each switching decision." That is a known problem for which hash-based direct addressing was a known, textbook solution — textbook KSR rationale (i).
• Predictable result, no change in principle. Replacing an O(n) linear scan of a URL table with an O(1) hash lookup is a predictable variation of a known technique, implemented with known hash-table machinery. KSR, 550 U.S. at 417.
• Design incentive: wire-speed switching was the industry's explicit design driver (Alteon marketed RISC-per-port WebICs for wire-speed L4–L7) — a POSITA had a concrete reason to eliminate per-packet table scans.
• Reasonable expectation of success: CARP and Cache Digests had already demonstrated that hashing a URL to a compact value → deterministic cache/state location worked at scale in the same decade.

Strength: high. Limitation (j) reads directly on a decades-old data-structure technique; limitation (i) is practiced by pre-2000 hash-routing; (a)–(h) are conceded or shipped art.

Combination 2 — § 103 ground for the F2 "prefetch based on hotness of sub-URLs" family (claims 1–42, [UNVERIFIED])

Primary refs: [ADMITTED-2] (Web switches "predict hot content") + HTTP/1.1 embedded-reference behavior [FIELD-ART-8] + Padmanabhan & Mogul predictive prefetching [FIELD-ART-9] + a hit-counter/hotness-ranking element as disclosed in the patent's own hot-URL table (itself a feature the Background concedes as "predict hot content").

Mapping. F2 requires: (i) parse returned content for embedded/sub-URL references; (ii) rank those sub-URLs by a hit counter; (iii) retrieve the highest-ranked sub-URL's content before the client requests it.

• (i) was inherent to any HTTP/1.1-aware proxy/browser — [FIELD-ART-8] and the patent's own admission that browsers "make new requests to retrieve the embedded references."
• (ii) hotness/popularity ranking is [ADMITTED-2] ("predict hot content") plus a hit counter, which the patent itself treats as routine bookkeeping.
• (iii) predictive prefetching of exactly the "next likely" object was the entire thesis of Padmanabhan & Mogul (1996) — retrieve before the request arrives, ranked by observed access pattern.

Motivation to combine. Solves the admitted, universally felt problem of "latencies caused by the Web's back-and-forth loading process" (patent's own words); both refs concern HTTP object retrieval; the combination needs no structural change to the switch/proxy — only adding a prefetch trigger keyed to a counter the switch already maintains. § 103 is "not confined to the predictable use of prior art elements according to their established functions" — here it is precisely that predictable use. KSR.

Counterpoint to be honest about: claim 43's digest/tag machinery is not needed for F2; but if a claim ties prefetch selection to the switch's hit counter, that is a two-reference combination (prefetch art + hot-content-tracking art) that is still weakly non-obvious only if one insists the counter must be switch-resident. I rate this moderate-to-high obviousness.

Combination 3 — § 103 ground for the F3 "cache content positioned by degree of hotness" family (claims 1–42, [UNVERIFIED])

Primary refs: [ADMITTED-2] ("replicate hot content to a Web cache") + popularity-based cache placement / LRU-LFU replacement, tiered caching, and geo-distributed caching (standard 1990s caching practice, incl. [FIELD-ART-6] CARP's distributed cache arrays).

Mapping. F3 requires (i) measuring degrees of hotness and (ii) storing/moving content to more-accessible positions or closer tiers based on that measure.

• (i) hotness measurement = hit counter + timestamp = [ADMITTED-2].
• (ii) "position content by popularity" is the canonical LFU/replacement-policy concept and the classic cache-hierarchy placement problem; tiered/regional caching for latency was standard CDN practice by 1999–2000. The patent's own summary phrases F3 as moving content "to a more accessible location … [or] in close proximity to the hot content" — i.e., a known solution to a known cache-management problem.

Motivation. Direct design incentive to maximize hit rate / minimize latency — KSR rational (iv). Adding a counter-driven reallocation trigger (reusing the very hot-URL table the switching function already builds) is a predictable, low-risk refinement.

Strength: moderate-to-high, especially against any claim that does not require a specific novel data structure.

Combination 4 — § 103 ground specifically for the § 112(f) system claim 52 and the SSL dependents (44–46, 54)

• Claim 52 is the means-plus-function mirror of claim 43; the same Combination 1 mapping applies to the disclosed structures (IFS 200 = parsing/forwarding means; cache processor 208 + digest generator 216 + cache 212 = "means for storing … at an address based on the digest value"). No independent § 103 work is needed — it rises/falls with claim 43.
• Claim 54 / 45 (decrypt cipher-text request): SSL/HTTPS offload acceleration was commercially standard by 1999–2000 and is directly in this art space — [FIELD-ART-4] (Apostolopoulos/Peris/Saha, Transport layer security: how much does it really cost?) quantifies the cost of serving secure pages and the fetch/replicate TLS handshake problem. Combining a known content switch with a known SSL offload front end to relieve origin servers is a textbook KSR predictable combination (Two known techniques; no functional interference).
• Claim 47 (reverse-DNS to enumerate IPs for a hot domain): routine DNS practice (RFC 1034/1035) applied to the admitted "hot server" concept.

6. Where a non-obviousness argument could survive — and why it is thin

Intellectual honesty requires identifying the patent's best defensive ground:

The strongest candidate limitation is the combination within one switch of (i) response-side tag-append on the outbound cookie and (ii) request-side digest-addressed state storage — i.e., the "tagging and digesting" pairing of claims 43/52. A patentee could argue:

• Prior art cookie-stickiness switches encoded the server identity into the cookie by rewriting/inserting the cookie, whereas claim 43 recites appending a tag to a field, and
• Prior-art hashing art (CARP/Ross) used hashing to select a cache, whereas claim 43 uses a digest to address a state store.

Why this is weak:

1. The patent's own advantage statement defeats it. The stated benefit is merely "more rapidly route … using tags with minimal processing" — a pure performance/optimization motivation, i.e., an affirmatively obvious reason to modify a known cookie-persistence switch.
2. "Tag appended to a field" is a labeling distinction, not a technical one. Whether the switch writes a server ID as the cookie value (ArrowPoint/Alteon) or appends it to a cookie field is a form choice with no new result — the epitome of a predictable variation. A POSITA would adopt whichever form parses faster, which is exactly the motivation the patent claims as its own.
3. Digest-to-select-cache vs. digest-to-address-state is the same primitive (hash → deterministic location) applied to two obviously interchangeable targets in the same system; using one hash to locate either content or its state record is a simple substitution of one known element for another. KSR.
4. No nexus to any objective indicia. There is no litigation, no IPR, no licensing, no industry praise, and no copying in the record (per the earlier sections). With the patent expired and never asserted, there is no secondary-considerations evidence to rebut the prima facie case.

Net: the patent's best argument would be a claim-construction argument (reading "tag" and "field" narrowly to exclude cookie-rewriting switches), not a § 103 argument. And because there is no PTAB or court construction on the books, that construction is untested either way.

7. Bottom line

The decisive single fact: the applicant admitted in the specification that prior-art Web switches already performed hot-content detection, hot-content replication to cache servers, sticky cookie connections, and content-to-server affinity. Under KSR, once the result is conceded as known, the remaining question is only whether the specific mechanism (small appended tag + hash-addressed state store) was a predictable design choice — and the specification's own "conventional flow switches … search a table … consuming substantial computational resources" passage supplies the motivating problem and its textbook solution, making the digest-addressing limitation squarely obvious. US 7,062,570 is, in my assessment, vulnerable to a strong § 103 challenge across essentially the entire claim set — but the caveats below materially qualify that view.

8. Caveats and the open verification items you should close

1. I did not have the patent's actual [56] reference list. I built this on the patent's self-admissions, two verified-public art-space papers (Apostolopoulos INFOCOM 2000; Ross IEEE Network 1997), the self-cited CARP, and commercially shipped ArrowPoint/Alteon cookie-persistence switches. Before filing anything, pull the verbatim 1449 block from https://patents.google.com/patent/[US7062570B2](/patent/US7062570B2)/en (Patent Citations section) or USPTO PatentCenter for 09/921,460.
2. Claims 1–42 remain [UNVERIFIED]. Earlier sections correctly flagged that verbatim text was retrievable only from claim 43. My F1/F2/F3 groupings are drawn from the abstract and SUMMARY and are inferences; do not quote them as claim language.
3. The two "not prior art" traps flagged: (a) the co-filed US 7,228,350 by the identical inventive entity is not § 102(e) art (see §3); and (b) per the earlier disambiguation table, the "’570" shorthand in certain PTACTS documents refers to other patents (e.g., the Parallel Networks '217 family) and must not be imported here.
4. Date discipline: the effective date is 2000-08-04 for matter carried in provisional 60/223,087; 2001-08-03 otherwise. Confirm each reference's public-availability date against the provisional content before relying on it — INFOCOM 2000 (March 2000), Ross (Nov/Dec 1997), CARP draft (Feb 1998), and RFC 2616 (June 1999) all clear the earlier date, but anything dated mid-2000 must be checked against the provisional's support.
5. This is an analytical assessment, not a certified invalidity opinion. A formal § 103 challenge would require the 1449 list, element-by-element claim charts against the granted claims 1–42, and (desirably) an expert declaration. Given the patent expired 2023-09-07 and has never been asserted or challenged, the practical need to build that record is low — but if a demand letter ever arrives citing 7,062,570, the § 286 lookback and the "claims stand as issued, never construed" posture in the earlier sections are the operative facts.