US 7512702

I'll search for this patent across the sources you specified.

Let me dig deeper on the claims text and check for any 2026 CAFC activity.

Search results summary

I ran targeted searches against Google Patents, uspto.report, FreePatentsOnline, and general queries aimed at USPTO/PatentCenter and CAFC 2026 dockets. Two important negative results up front:

• "CAFC 2026 docket 7512702" → returned no results. No evidence of any 2026 Federal Circuit activity tied to this patent number.
• "USPTO PatentCenter US7512702B1" → returned no results via the search interface.

Most of my search hits were cross-citations to US7512702B1 in other patents' reference lists, not the patent record itself. The one substantive record I retrieved was uspto.report/patent/grant/7512702, plus the authoritative Google Patents text supplied in your message. I did not obtain a verbatim copy of the full claim set, so claim-level descriptions below are flagged accordingly.

Bibliographic data (from the authoritative Google Patents record)

Relationship note: This is a continuation-in-part of Ser. No. 10/102,287 (filed Mar. 19, 2002, issued as US 7,047,315). Domestic priority under 35 U.S.C. §120 is claimed in the specification.

Abstract (as recorded)

"A method and apparatus providing highly scalable server load balancing are disclosed. Data packets from a client are routed through one or more routers to a server load balancer, which is selected from among a plurality of server load balancers in a network. In response to receiving a request packet, a particular server site to process the client request is selected. A first path to a second router associated with the particular server site, and a second path to a server load-balancing device associated with the second router, are determined. A mapping of flow identifying information, associated with the packet, to a first label value that identifies the first path and to a second label value that identifies the second path, is created. The first label value and the second label value are stored in the packet. All subsequent packets associated with the client request are forwarded to the server load-balancing device based on looking up the first label value and second label value in the mapping. As a result, a network is scalable to process and load-balance numerous client requests, which are efficiently routed to the site, server load-balancer, and server that are handling the request."

Plain-language overview of the independent claims

⚠️ Uncertainty flag: The full patent text supplied to me contains the abstract, description, and figure listings, but not the verbatim claim section. I could not retrieve the complete claim set from the search results either. The following is reconstructed from the abstract, the "Summary of the Invention," and one confirmed claim fragment from uspto.report — it is directionally reliable but not a substitute for the literal claim language. I cannot state the exact count or exact numbering of independent claims with high confidence.

Confirmed claim fragment (from uspto.report): a claim recites "forwarding the packets to the server load-balancing device based on the first MPLS label value and second MPLS label value" — confirming that claim 1 recites MPLS label values specifically (not merely generic "label values").

Claim 1 (method) — likely structure: A method of load balancing in which a router receiving a client request packet: (1) selects a server site to handle the request; (2) determines a first path through the network to a second router associated with that site, and a second path to a server load-balancing device associated with that second router; (3) creates a mapping from flow-identifying information of the packet to two MPLS label values — one identifying each path; (4) inserts both label values into the packet; and (5) forwards subsequent packets of the same flow to the load-balancing device by looking up the labels in the mapping. The problem it solves: it avoids re-running global server-load-balancing decisions on every packet, and avoids TCP termination and per-flow state tables (5-tuple tracking) at the load balancer.

Additional independent claims — per the "Summary of the Invention": The specification states the invention "encompasses a computer apparatus, a computer readable medium, and a carrier wave configured to carry out the foregoing steps." This indicates roughly three further independent claim types beyond the method claim:

• Apparatus claim — a router/network device (processor + memory with instructions) configured to perform the select-determine-map-store-forward steps.
• Computer-readable-medium claim — instructions that, when executed, cause the above steps.
• Carrier-wave claim — a propagated signal carrying the instructions.

Two-level label stack: A recurring independent-claim concept is the "two-level stack" — an upper-most label directing the packet to the last content router (L-CR) and a lower-most label providing "client stickiness" to the specific server/SLB device. The specification also notes the two-level stacking process of RFC 2547 may be used, and that the upper-most label may instead be IP tunnel encapsulation.

Litigation / CAFC 2026 status

• I found no CAFC 2026 docket involving US 7,512,702. Searches returned empty.
• Consistent with that, the patent is recorded as Expired – Lifetime (adjusted expiration 2024-02-17), which makes live 2026 appellate litigation unlikely, though an appeal arising from pre-expiry district court litigation is not impossible.
• Caveat: My negative result is a search negative, not a confirmation of absence. I could not query the CAFC docket system or USPTO PatentCenter directly. If you need a definitive answer, check PACER/CM-ECF and the CAFC docket directly, and confirm the expiration/adjustment details via USPTO PatentCenter.

Where I am uncertain

1. Exact claim text and independent claim count/numbering — not retrieved. Do not rely on my claim overview as literal language.
2. Whether any 2026 CAFC appeal exists — no results found, but not definitively ruled out.
3. The adjusted expiration date of 2024-02-17 — reported as listed by Google Patents; the reason (PTA, terminal disclaimer, or a listed calculation) is not stated in the record I retrieved.
4. The Google Patents record conflates "prior art date" (2002-03-19) with the actual filing date (2002-07-19); I have kept both distinct above rather than auto-correcting either.

If you want, I can attempt a more targeted retrieval of the verbatim claims from the USPTO full-text (patft) or the Google Patents claims section specifically.

I'll research litigation involving US patent 7512702 across patent litigation databases.

Answer: No litigation involving US 7,512,702 was found

Based on my searches (Google Patents full text, general web search for litigation mentions, and litigation-database results surfaced via Unified Patents–type aggregators, Patsnap litigation records, Justia/CourtListener dockets, and SEC filings referencing Cisco patent suits), I could not identify any lawsuit — district court, ITC, or CAFC appeal — in which U.S. Patent No. 7,512,702 was asserted or adjudicated.

I want to be explicit about the limits of this answer rather than imply a negative I haven't fully proven:

What I checked and what it returned

1. Patent identifiers. Every "7512702" hit that appeared was a citation in another patent's reference list, not a litigation record. Example: the front matter of U.S. Pat. No. 8,936,552 (NEC) and U.S. Pat. No. 7,676,599 (i2 Telecom IP Holdings) lists "7512702 | Method and apparatus providing highly scalable server load balancing | 2009-03-31 | Srivastava et al." That is the correct patent — title and inventors match the Google Patents record you provided (Sunil K. Srivastava, Daniel C. Tappan; assigned to Cisco Technology, Inc.).

2. Cisco enforcement campaigns. I surfaced a number of Cisco patent cases, but in each the patents-in-suit are different numbers:

   • XR Communications, LLC d/b/a Vivato Technologies v. Cisco Systems, Inc., C.D. Cal. No. 2:17-cv-02951 (patents 7,062,296; 7,729,728; 6,611,231).
   • Corrigent Corp. v. Cisco Systems, Inc., W.D. Tex. No. 6:22-cv-00396 (patents 6,957,369; 7,113,485).
   • NetSocket, Inc. v. Cisco Systems, Inc., E.D. Tex. No. 2:22-cv-00172 (NetSocket-owned patents 7,606,885; 7,734,796; 7,616,601; 7,720,966; 7,190,698; 7,827,284; 7,885,286).
   • SRI Int'l v. Cisco (D. Del.), Paltalk Holdings v. Cisco (W.D. Tex. — patent 6,683,858), Enhanced Data Streaming LLC v. Cisco (E.D. Tex. No. 2:25-cv-01038 — patents 9,185,151/RE50,398; 6,891,855; 7,336,605; 7,660,234; 7,032,135).
   • The Cisco v. Arista ITC/district court matters referenced in Arista's SEC filings (patents such as 7,162,537; 6,741,592; 7,200,145; 7,340,597; 7,023,853; 7,224,668; 8,356,296). I did not find 7,512,702 among those.

   None of these involve 7,512,702.

3. Procedural context for the patent. The Google Patents bibliographic record shows the application (Ser. No. 10/199,760, filed 2002-07-19, priority claimed from Ser. No. 10/102,287 filed 2002-03-19, now U.S. Pat. No. 7,047,315) and legal status "Expired – Lifetime, expires 2024-02-17." So the patent's enforceable term ended in early 2024. That is consistent with (though not proof of) the absence of recent assertions.

Important caveats

• I could not complete a direct query of Unified Patents' litigation caselist or PACER/Docket Navigator for the literal number — a tool call limit was reached, and the results returned were for other patents (as noted above, several similarly numbered Cisco patents appear in Cisco's campaigns, which is exactly the kind of look-alike you warned against). A search for "7512702" combined with litigation terms returned no results at all.
• Because 7,512,702 is Cisco-owned, any suit would have Cisco (or a Cisco successor/assignee) as plaintiff-assertor, not defendant. My searches found no Cisco complaint asserting it.
• Aggregator databases (Unified Patents, RPX, Patsnap, Docket Navigator) are not exhaustive; settled cases dismissed early, ITC investigations, and state-court actions can be under-indexed.

Bottom line

On the evidence available to me, I found no known litigation involving U.S. Patent No. 7,512,702. I am reporting this as "not found," not as a verified negative. If you need a legally defensible clearance statement, the definitive check is a direct docket query (Docket Navigator or Lex Machina) on patent number 7,512,702, plus a PACER/CAFC docket search for the patent's title ("Method and apparatus providing highly scalable server load balancing") and its assignee, Cisco Technology, Inc. I could not complete those queries within this session and do not want to represent otherwise.

I'll verify the default "no PTAB activity" position against live sources before writing anything up.

Let me run more targeted checks — including the assignee/technology angle and any Unified Patents or Cisco-related challenges.

Proceedings overview

Zero AIA trial proceedings are on file against US 7,512,702. The canonical structured source for this task (USPTO Open Data Portal "PTAB proceedings on file" block) returns no IPR, PGR, or CBM proceeding for this patent, and my independent web checks surfaced no petition, institution decision, final written decision, or Federal Circuit appeal naming US 7,512,702 — so the breakdown is 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution-denied. The defensive posture this gives a defendant is not "hardened" and not "already killed" — it is untested at the PTAB, but time-expired: per the patent record, the legal status is Expired – Lifetime, adjusted expiration 2024-02-17, which means the challenge math is driven less by PTAB strategy than by the § 286 damages lookback and whether any live assertion exists.

Verification note (what I did and did not confirm): I searched for PTAB/PTAB-related filings, petitions, FWDs, appeals, and reexamination requests tied to "7512702"/"7,512,702," including the Cisco/load-balancing context and aggregator (Unified Patents) activity. The only substantive record that surfaced was the grant page at uspto.report/patent/grant/7512702 and the Google Patents record at patents.google.com/patent/US7512702. I found no evidence of a PTAB trial, no CAFC docket, and no litigation hit. I cannot rule out a pre-AIA inter partes reexamination or ex parte reexamination from search alone — the ODP block covers AIA trials, not reexams — so that is the one gap worth closing via Patent Center before you rely on "no administrative history at all." Treat the above as "no AIA trial activity," not "no validity challenge ever."

Proceedings (by proceeding)

There are no proceedings to enumerate. The per-proceeding template (Type / Filed / Status / Judge panel / Petition grounds / Institution decision / FWD / Settlement / Appeal / Defensive value) is therefore inapplicable, and I will not manufacture a proceeding number, panel, or claim-level disposition to fill it. Any such entry would be fabrication.

For completeness, here is what does exist on the record for this patent, drawn from the authoritative patent text:

Strategic summary

Canceled vs. sustained vs. untested. No claim of US 7,512,702 has ever been adjudicated unpatentable by the PTAB, and no claim has been sustained against a PTAB challenge either. Every claim is untested at the PTAB. There is no FWD to quote, no cancellation to point at, and no claim-level "claim 1 is dead" argument available to a defendant. Contrast that with the patent's own description of prior art practice on flow-state tracking and TCP termination — the specification spends substantial text attacking the TCP-termination/NAT state-tracking model as a choke point, which is exactly the framing a future petitioner would invert against the claims. That attack has simply never been filed.

Estoppel landscape — § 315(e)(2). Because no one has petitioned, no § 315(e)(2) estoppel attaches to anyone. No petitioner, no privies, no IPR estoppel, no real-party-in-interest chain to worry about. A defendant today is not blocked from raising any § 102/§ 103 ground in an IPR, and the § 315(b) one-year clock only starts if and when a complaint is served. Practically, the estoppel question is moot: the relevant limitations are temporal, not estoppel-based. PGR is unavailable (the 9-month post-grant window closed long ago — the patent issued 2009-03-31). CBM review is unavailable to a new petitioner (the transitional program sunset; new CBM petitions are not accepted). That leaves IPR as the only AIA vehicle, and it can only reach § 102/§ 103 art — no § 112 or § 101 theories.

Pattern signals. There is no petitioner to look for a repeat filer, no aggregator in the chain (no Unified Patents filing was found), and no patent-owner PTAB appeal activity because there was never a PTAB trial to appeal. The absence is itself informative and fits the profile: the patent is Cisco-owned, not NPE-owned, and Cisco patents are rarely the target of the aggregator community — Unified Patents and similar entities overwhelmingly target asserted NPE patents, not operating-company portfolios. A Cisco-owned patent that nobody is asserting is not an IPR magnet. The more important pattern signal is the expiration date: an expired patent cannot ground an injunction or ongoing royalty, and IPR against an expired patent is permitted but of sharply diminished value (the patent owner cannot amend claims, the Board construes under Phillips, and the sole practical win is cancellation of claims that bear only on past damages). Expect low petition interest absent a live damages case.

Recommended next steps

1. If you are a defendant / accused infringer: there is no FWD to link and no canceled claim to lean on — do not tell a court or an adversary that this patent "has been invalidated at the PTAB." It has not. Your first move is to determine whether the assertion is even live in a damages-relevant window. The patent's record status is Expired – Lifetime, expires 2024-02-17 (Google Patents; verify against USPTO Patent Center). Expiration caps prospective liability; recovery, if any, is limited to infringement within the six years preceding filing under 35 U.S.C. § 286, and the demand/assertion must identify pre-expiration activity.
2. Close the reexam gap before disclaiming all administrative history. The ODP block covers AIA trials only. Check PTAB E2E at https://ptab.uspto.gov and USPTO Patent Center for (a) any pre-AIA inter partes reexamination, (b) any ex parte reexamination or reexamination certificate, and (c) any reexam or certificate that canceled or confirmed claims — a reexam certificate would be claim-level prior art-bounded narrowing and would matter more than an IPR here. Also confirm the § 286/expiration computation against the face of the patent (PTA/terminal disclaimer), since the 2024-02-17 date is an assumption.
3. If you are contemplating a defensive IPR: know the constraints up front — IPR is the only available AIA route (§ 102/§ 103 only; PGR window long closed; CBM program unavailable); no § 315(e)(2) estoppel binds you or any prior petitioner because none exists; and the § 315(b) one-year clock begins on service of a complaint asserting the patent. Weigh the cost against the fact that the patent is expired — the realistic objective is eliminating a past-damages claim, not clearing a royalty for future shipments.
4. Litigation/appeal check. Searches for district court assertions and a Federal Circuit docket came back empty. Confirm independently via CourtListener (https://www.courtlistener.com) and PACER before representing to anyone that the patent has never been litigated or appealed — a negative search result is not the same as a docket review.

Bottom line: the honest answer is no PTAB activity on file — no petitioner has ever attacked US 7,512,702, so there is no invalidation outcome, no claim-level narrative, and no estoppel to exploit. The patent's defensive significance today flows from its expired status (adjusted expiration 2024-02-17) rather than from any PTAB result, and the one item that could change this picture is a reexamination or reexamination certificate that the AIA-trial feed would not show.

I'll research the assignment record for US 7,512,702 across USPTO Assignment Center, Google Patents legal events, and related sources.

Let me look for the specific reel/frame and any post-issuance transfers.

Assignment & Ownership Analysis — US 7,512,702

Scope note / evidentiary limitation (stated up front): I was unable to complete a direct query of USPTO Assignment Center (assignmentcenter.uspto.gov) or a per-record retrieval of the reel/frame numbers for this patent within this session; my fetches against Google Patents legal events and uspto.report did not expose the reel/frame on the rendered pages. I did not — and will not — supply a reel/frame number I could not verify. What follows is built from the authoritative Google Patents bibliographic/legal-events record and the issued-patent front matter (the US7512702B1 PDF), which do confirm the existence and parties of the assignments below.

Inventors

Both are named on the issued patent front matter (75) as inventors, and both are listed in the Google Patents reassignment record as the assignors to Cisco Technology, Inc. executed/recorded 2002-07-19 — the same date as the application filing. A same-day inventor-to-employer assignment is the standard signature of an employee invention-assignment obligation, so Cisco employment at filing is a reasonable inference.

Unusual patterns: None observable. I found no evidence of either inventor departing Cisco within 12 months of filing, no subsequent inventor-side assignments, and no reissue/reexamination that would indicate abandonment. I cannot verify departures (inventor employment histories are not in the assignment record); treat "no departure" as "not observed," not as a proven negative.

Original assignee

Cisco Technology, Inc., San Jose, CA — the assignee on the face of the issued patent (73) and the assignee of record on the 2002-07-19 assignment.

• Business: Networking hardware/software (routers, switches, load-balancing products — see the "Distributed Director" and "Local Director" commercial references in the specification's Background). The patent is a Cisco server-load-balancing invention.
• Product embodying the claims: Cisco's family of server load-balancing products (e.g., Cisco Content Switching / Local Director / Distributed Director lineage, and MPLS-capable IOS routing platforms) is the commercial context in which the claims would be practiced. I did not find a Cisco datasheet or 10-K line item expressly mapping the 7,512,702 claims (2-level MPLS label stack mapping flow-identifying info to an F-CR→L-CR path and an F-CR→SLB path) to a named SKU. Treat "ships a product embodying the claims" as likely but not documentary-confirmed.
• Current status: Operating, public company (NASDAQ: CSCO). No bankruptcy, no acquisition, no change of name reflected in the record. Google Patents lists Current Assignee: Cisco Technology, Inc. and the legal status as "Expired – Lifetime, expires 2024-02-17" — i.e., the patent ran its full term to expiry still held by the original operating assignee.

Assignment timeline

The recorded chain is a single link: inventor → Cisco, recorded contemporaneously with filing. Google Patents shows exactly one reassignment event, and no further transfer events on the legal-events timeline.

• 2002-07-19 (executed) / recorded 2002-07-19 — Reel/Frame not retrievable in this session (USPTO Assignment Center query not completed; do not rely on an unverified number)
  • Conveyance: Assignment (recorded as "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: Sunil K. Srivastava; Daniel C. Tappan
  • Assignee: Cisco Technology, Inc.
  • Correspondent: Not confirmed. I could not retrieve the recorded correspondent field. (Cisco's 2002-era recordings were filed through Cisco's in-house patent department; I am flagging this as unverified rather than naming an attorney or firm I cannot cite.) — No recurrence can be assessed at this time.
  • Context: Standard employee→employer invention assignment; executed on the filing date of application 10/199,760.

Related family (context, not a separate link in this patent's chain): 7,512,702 is a continuation-in-part of Ser. No. 10/102,287, filed 2002-03-19, now US 7,047,315 (per the cross-reference in the specification and Google Patents' external-priority link). That parent is in the same Cisco-originated family, so any ownership-chain query for the portfolio should include 7,047,315 as well — it also starts and (presumably) ends at Cisco.

No post-issuance assignments, no security interests, no mergers, no releases, and no transfers to any third party were found. Absent a transfer, USPTO records would still name Cisco Technology, Inc. as sole owner of record — consistent with the Google Patents "Current Assignee" field.

Timeline diagram

timeline
    title Ownership of US 7512702
    2002 : CIP filed by Srivastava and Tappan
         : Assignment recorded to Cisco Technology Inc
    2009 : Patent issued Mar 31
    2024 : Patent term expired Feb 17

NPE / troll-pattern signals

Cross-reference consistency check: The prior litigation section concluded no litigation names 7,512,702 and that the patent expired 2024-02-17. Both are consistent with everything above — no transfer, no assertion, full-term expiry in Cisco's hands. No contradiction found.

Verdict

Insufficient data (no records, or only the original assignment).

Justification: the record contains exactly one recorded conveyance — the 2002-07-19 inventor-to-Cisco assignment executed on the filing date of application 10/199,760 — with no subsequent assignments, no security interests, and no changes of name. The chain therefore terminates at an operating company (Cisco Technology, Inc.), which defeats every NPE signal in the checklist; "Insufficient data" is selected here strictly per the rubric's literal catch-all for "only the original assignment," not because ownership is unclear — on the evidence, the affirmative finding is a non-NPE, operating-company ownership profile. Picking any NPE category would require an unverified reel/frame or an unconfirmed correspondent name, which I decline to fabricate.

Verify at: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search by patent number 7512702); mirrored bibliographic record at Google Patents — https://patents.google.com/patent/US7512702/en; issued front matter confirming assignee/inventors at US7512702B1 PDF.

Open items for a definitive clearance run: (a) retrieve the reel/frame and correspondent of record for the single 2002 assignment from Assignment Center; (b) confirm whether the C-I-P parent US 7,047,315 carried any post-issuance transfer (none expected, but the family should be swept together); (c) confirm no assignment was recorded against the application number 10/199,760 that never propagated to the issued patent's legal-events list.

I'll research the patent citations and prior art references for US 7,512,702.

Let me get the actual "References Cited" (prior art) section — the citations ON the patent, not the forward citations to it.

I hit a tool-step limit mid-research, so let me be transparent about exactly what I retrieved and what I could not, then give you the best-grounded prior-art analysis I can.

⚠️ Critical limitation on this section

I could not retrieve the verbatim "References Cited" front-page section of US 7,512,702 itself. The authoritative full text you supplied (Google Patents) contains the abstract, description, and drawings — but not the front-page patent-citation block. My web searches returned almost entirely forward citations (patents that cite 7,512,702 — which by definition are later art and therefore not §102 prior art to it) plus family members and co-field citation lists. A direct fetch of the "Patent Citations / Cited By" subsections of patents.google.com/patent/US7512702/en was cut off by the step limit.

Therefore: I cannot give you a verified, item-by-item list of the references cited of record in US 7,512,702, and I will not fabricate one. Everything below is tiered by confidence, and I flag precisely where confidence is low.

A. What the record does reliably show

Target patent (unchanged from prior sections): US 7,512,702 B1 — "Method and apparatus providing highly scalable server load balancing"; inventors Sunil K. Srivastava, Daniel C. Tappan; assignee Cisco Technology, Inc.; App. No. US 10/199,760; filed 2002-07-19; priority 2002-03-19; issued 2009-03-31.

Family members (co-owned, same inventors → NOT §102 prior art):

These share inventors/priority with 7,512,702 and are Cisco-owned, so they are family, not anticipatory art — but they are the single most conceptually overlapping documents in the space.

B. Prior art named inside the patent's own Background (self-identified art)

These are the systems/standards the specification itself describes as the pre-existing state of the art. They are real and citable, but note: naming something in the Background does not constitute an admission that it is prior art (the patent expressly disclaims this — "the approaches described in this section… are not admitted to be prior art"). Dates below are as given or general knowledge.

Confidence: high that these are the art discussed in the specification; low that they are the references cited of record (see §C).

C. Patent references appearing in the co-field citation lists I retrieved (candidate §102 art — UNVERIFIED as cited-of-record in 7,512,702)

The searches surfaced the front-matter reference lists of neighboring patents — US 8,930,552 (NEC, granted 2015-01-06) and US 8,090,877 (Citrix, granted 2012-01-03) — both of which list US 7,512,702 among their references. Those lists also contain the classic SLB/TCP-splicing/MPLS references of this field. These are the most likely members of 7,512,702's own cited-art set, but I have NOT confirmed that 7,512,702 cites them. Treat the §102 column as screening hypotheses.

Timing caution: several items above publish after the 2002-03-19 priority date. Under §102 they would only qualify via an earlier effective filing date. I am flagging this rather than asserting they are art — I have not verified their filing/priority dates.

D. Forward citations (NOT prior art — listed for completeness)

The following cite 7,512,702 and therefore postdate it; they are not §102 art but confirm the technology's downstream footprint:

• US 8,930,552 (NEC) — application switch routing
• US 8,090,877 (Citrix) — cookie proxying
• US 8,736,274 / US 7,676,599 and others listing "7512702 … Srivastava et al." in reference lists
• US 8,206,416; US 12,047,441; US 8,908,700; US 9,826,033; US 2014/0173045 (various — all cite it as a reference)

E. §102 anticipation assessment (preliminary, and I must stress "preliminary")

No single reference I can identify discloses every element of reconstructed claim 1, which requires all of: (i) selecting a server site; (ii) determining a first path to a second router AND a second path to an SLB device; (iii) creating a mapping of flow-identifying info to two MPLS label values; (iv) storing both labels in the packet; (v) forwarding subsequent packets by looking up the labels. Anticipation under §102 requires all elements in one reference.

• Colby (US 6,006,264 / 6,449,647): potentially meets the flow-identification → selected server element and directed-flow forwarding, but no MPLS two-level label-in-packet mechanism → §103 more likely than §102.
• Bhagwat (US 5,941,988) / Hensbergen (US 7,000,027): address TCP splicing but not the label-mapping architecture → not anticipatory; possible §103.
• The family (US 7,047,315; US 2005/0149531 A1): same inventors/priority → cannot be §102 art for 7,512,702's claims (double-patenting/§102(b) same-inventor issues aside).
• The two-level MPLS stack of RFC 2547 is a standard, not a single anticipatory reference to the claim as a whole.

F. What you should verify before relying on this

1. The actual cited-reference block. Pull it directly from the USPTO full-text (PatentCenter/PatFT) or the Google Patents front page for US 7,512,702 — that is the only authoritative source for "each patent citation for 7512702." I could not reach it.
2. Filing/priority dates of the timing-suspect publications in §C to establish whether they qualify as §102 art.
3. Whether the non-patent literature (WCCP draft draft-forster-web-pro-00.txt, RFC 2547, RFC 2991, distributed-Director/DRP documentation) was cited of record, since the specification leans heavily on those as background.

Bottom line: I could name and date the family, the self-described background art, and the co-field patent references that cluster around 7,512,702, and give a screening §102/§103 read on each. But I do not have high confidence in the complete, verbatim "References Cited" list for 7,512,702, and the strongest single on-point candidates (Colby) point more toward §103 than §102. I would not represent the §C table as the patent's confirmed citation record without the direct PatentCenter lookup.

I'll start by flagging a discrepancy and then gathering grounding for the prior-art analysis.

⚠️ Discrepancy flag: The current task says "Use the results from the Prior Art section of this page," but no Prior Art section was included in the previously-generated sections supplied to me (I have only the Patent summary and Litigation summary). I will not invent a prior-art section. Instead I'll ground the analysis in (a) the prior art cited on the face of the patent itself and (b) the standards/document prior art the specification expressly discusses, and I'll verify what I can via search.

Good grounding. Now let me try to retrieve the actual claim language and the front-page "References Cited" list, which I flagged as missing in the earlier summary.

I've now retrieved the verbatim claim 1, which materially changes the analysis. Here is the § 103 obviousness assessment.

0. Two mandated corrections to the earlier sections

⚠️ Correction #1 (contradicts the earlier Patent summary). The earlier summary stated I could not retrieve the claim text and reconstructed claim 1 as a "select → determine two paths → map to two MPLS labels → forward" method. I have now retrieved the literal claim 1 (from RPX Insight, https://insight.rpxcorp.com/patent/[US7512702B1](/patent/US7512702B1)). The earlier reconstruction was directionally right but materially incomplete — it omitted three limitations that are likely the whole ballgame for § 103: (a) label distribution via a discovery request packet, (b) the negative limitation that the SLB device does not terminate any TCP connections, and (c) a cookie containing server address information. Those omissions would have produced a too-broad (i.e., too easy to invalidate) claim reading.

⚠️ Correction #2. The task instructs me to "use the results from the Prior Art section of this page." No such section was supplied to me (I have only the Patent summary and Litigation summary). I will not fabricate one. Instead I ground the analysis in (i) the references the patent itself cites and discusses, and (ii) references I independently verified this session, with URLs. If you can paste the actual Prior Art section, I'll re-run the chart against it.

1. Governing law, dates, and the person of ordinary skill (POSITA)

• Pre-AIA law applies. Filed 2002-07-19; earliest priority 2002-03-19. So pre-AIA §§ 102/103 govern, with KSR Int'l v. Teleflex, 550 U.S. 398 (2007) supplying the obviousness gloss.
• Critical date for § 102(b): March 19, 2001 (one year before priority; July 19, 2001 if you use the actual filing date). Anything published before that is § 102(b) art regardless of common ownership or "by another" — so Cisco-authored publications like RFC 2547 are not disqualified by pre-AIA § 103(c).
• POSITA: a B.S. in CS/EE (or equivalent) plus ~2–4 years designing/operating IP data networks, with working familiarity with TCP/IP, L2/L3 load balancing (NAT/Direct-Routing/Local Director-style rewriting), MPLS label switching, and DNS/HTTP-based global server selection. (Nothing in the record defines the level of skill; this is a reasoned construct, flagged as such.)

2. Claim 1 (verbatim) and element-by-element chart

1. A method of providing highly scalable network server load balancing, the method comprising the steps of:

• receiving, at a first router in a path from a client to one or more server sites, a packet of a request from a client;
• selecting a particular server site to process the client request;
• determining a first path from the first router through a network to a second router associated with the particular server site, and a second path from the first router to a server load-balancing device associated with the second router;
• creating and storing a mapping of flow identifying information, associated with the packet, to a first multi-protocol label switching (MPLS) label value that identifies the first path and to a second MPLS label value that identifies the second path;
• storing the first MPLS label value and the second MPLS label value in the packet;
• distributing the first and second MPLS label values to the network by attaching the first and second MPLS label values to a discovery request packet that is sent into the network by the first router;
• forwarding all subsequent packets associated with the client request to the server load-balancing device based on looking up the first MPLS label value and second MPLS label value in the mapping, wherein the server load-balancing device does not terminate any TCP connections;
• receiving a cookie containing server address information from the second router, wherein the server address information identifies a server that is managed by the load-balancing device; and
• storing the server address information in the mapping of flow identifying information.

Note on the Background disclaimer: the '702 Background carries the boilerplate "the approaches described in this section … are not prior art." So I do not rely on the Background as a § 102(b) admission. I rely on the underlying public documents (RFC 2547, the WCCP draft, Distributed Director/Local Director product literature, US 6,006,264) which are prior art independently of the patent's characterization.

3. Combinations that would render claim 1 obvious

Combination A (primary): Colby '264 + RFC 2547 + WCCP/discovery-based label distribution

Colby '264 supplies elements 1a, 1b, 1d, 1g entirely (content-aware flow switch; site/server selection on proximity, load, congestion; per-flow forwarding). RFC 2547 supplies 1c and 1e, and does so for exactly the reason the '702 patent gives for its own invention — RFC 2547 states: "it is the two-level labeling that makes it possible to keep all the VPN routes out of the P routers, and this in turn is crucial to ensuring the scalability of the model." That is the same scalability rationale as the '702 title and abstract. WCCP-style discovery supplies 1f, the vehicle for distributing the labels.

Motivation: (i) identical field (IP forwarding, content routing, scalability of core routers); (ii) RFC 2547's express scalability rationale matches the '702's stated problem (SLB choke point); (iii) predictable result — substituting label-switching for hop-by-hop forwarding is a known, routine optimization (RFC 3031); (iv) KSR: "using a known technique to improve similar devices in the same way."

Combination B: Distributed Director/DRP (global SLB) + Colby '264 (local SLB) + MPLS label stacking

Maps the claim's two-tier architecture directly: DD selects the site/FQDN (1b); Colby's flow switch selects the server within the site (1a/1g); MPLS two-level stacking (RFC 2547/3032) provides 1c/1e. Motivation: the '702 patent's own architecture mirrors the known global/local split of Distributed Director + Local Director — combining a known global LB with a known local LB is precisely the architectural layering a POSITA would pursue for scale, and MPLS is the known tool for eliminating per-hop route lookups in the core.

Combination C: Anycast-based server selection + MPLS two-level stacking

Elements 1a/1b (an "Anycast node" selecting a server and forwarding toward a site) are the standard Anycast server-selection model described throughout the '702 specification (nodes A–E, First-CR/Last-CR). Combined with RFC 2547's two-level stack for 1c/1e (top label → Last-CR, bottom label → site edge/SLB), this is a near-complete reading of 1a–1g. Motivation: Anycast inherently needs an "is the site reachable and best?" decision at each node; MPLS is the known way to fast-switch the rest of the flow without repeating that decision.

Combination D: (A, B, or C) + non-terminating NAT/Direct-Routing LB + cookie affinity

Needed only for 1h and 1i. Local Director-style L3 rewriting (destination-IP rewrite) and Direct-Routing/IP-tunneling LBs do not terminate TCP — satisfying the negative limitation. Cookie-based affinity is a well-known, low-effort mechanism (the '702 background concedes it). Motivation: the Background of '702 articulates the known deficiencies of terminating TCP in a device "just in front of a Web server farm in a high-speed, high-volume network" — i.e., the art already recognized the problem, and the known Direct-Routing/NAT alternatives were the obvious response, combined with cookies for persistence, which Colby '264's own claim 37 already contemplated ("persistent connectivity").

4. Why a POSITA would have combined these (articulated, per KSR factors)

1. Same field, same problem. Every reference is in IP network load balancing / label switching and addresses the identical concern: the LB or core router becoming a stateful choke point.
2. Predictable use of prior-art elements. RFC 2547 already teaches two-level MPLS stacks where the bottom label identifies the destination edge device and the top label tunnels the core — a small, foreseeable step to re-purpose as "top = site/L-CR, bottom = SLB device."
3. Design incentive / recognized need. The '702 Background (and RFC 2547's §11 scalability section) independently document the commercial pressure to keep intermediate nodes stateless.
4. Known label-distribution vehicles. WCCP-style discovery, LDP/TDP (RFC 3036), RSVP-TE, and BGP/OSPF extensions were all known ways to move labels; the '702 spec itself lists SYN-packet piggybacking and Content Discovery Protocol as interchangeable alternatives — evidence the choice among them was design choice, not invention.
5. Known alternative to TCP termination. NAT/L2/L3 rewriting LB (Local Director) had long existed and avoids termination; the negative limitation 1h is therefore met by well-known operation.

5. Where the obviousness theory is weakest (and I should be candid)

1. "Discovery request packet" (1f). Neither Colby '264 nor RFC 2547 teaches distributing per-flow MPLS labels by attaching them to a discovery packet. The strongest art here is the WCCP draft the '702 Background cites — but I could not verify that draft's contents this session, and the Background's own boilerplate disclaims it as prior art. This is the limitation most likely to have carried allowance.
2. Granularity mismatch. RFC 2547 maps routes/VPN-IPv4 prefixes to labels (aggregate), whereas claim 1 maps flow identifying information (per-flow 5-tuple) to two labels. A POSITA could reasonably say this is not a mere substitution but a re-purposing of MPLS from aggregate forwarding to per-flow load balancing — and the '702 Background expressly criticizes per-flow 5-tuple state tables as unscalable. This tension (claiming a flow-state mapping while touting "without keeping flow states") cuts both ways and is worth checking: 2/3/4 of the "Prior art keywords" on Google Patents are server, router, packet, client, load, with "flow" logic threaded through — consistent with a per-flow-state claim.
3. The cookie + no-TCP-termination pairing (1h+1i). If the SLB device does not terminate TCP, the mechanism by which the second router returns a cookie into the flow is non-obvious from the face of the claim; this could raise a written-description/enablement question independent of § 103 (and would be a better attack vector than obviousness for this element).
4. File history is missing. I do not have the front-page "References Cited" list, the examiner's art, or the applicant's remarks. Claim 1's unusual combination of limitations (discovery-packet label distribution + no-TCP-termination + cookie) strongly suggests the applicant traversed art on those grounds. Any defensible § 103 opinion requires the file wrapper.

6. Other claims

• Independent claim 10 (computer-readable storage medium), claim 11 (apparatus, means-plus-function mirroring claim 1), and claim 12 (load-balancing router) track claim 1's limitations; they rise or fall with it. Note claim 11 is means-plus-function, so its scope under § 112(f) is bounded by the corresponding structure in the '702 specification — relevant both to validity and to any reading-on analysis.
• Dependent claims 13–22 — ⚠️ I have not retrieved their text. Under KSR, dependent claims adding a "two-level stack," RFC 2547-style VPN two-level stacking, or "upper label = L-CR / lower label = stickiness" would be obvious over Combination A joined with RFC 2547's express two-level teaching (the '702 spec itself cites RFC 2547 for the two-level stack). Do not treat this as a claim-by-claim opinion until the dependent claims are pulled.

7. Secondary considerations

• The earlier Litigation summary found no litigation involving US 7,512,702 — so there is no adjudicated nexus to commercial success, no jury finding, and no secondary-considerations record.
• Status is Expired – Lifetime, adjusted expiration 2024-02-17, so any § 103 analysis today is retrospective (back-damages or defensive). No CAFC 2026 docket was found (per the earlier litigation section — a search negative, not a confirmed absence).

8. Bottom line

Claim 1 as issued is a close call, and the answer turns almost entirely on element 1f. On the art I could verify, Colby '264 alone supplies most of the method (router-side interception, site/server selection, per-flow forwarding), and RFC 2547 supplies the two-level MPLS label stack with an express, identical scalability rationale — producing a strong prima facie case of obviousness on elements 1a–1e and 1g, and a defensible case on 1h/1i via Local Director-style non-terminating rewriting and known cookie affinity. The elements least likely to be rendered obvious are (i) distributing per-flow MPLS labels specifically on a discovery request packet, and (ii) the internally tense "no TCP termination" + cookie pairing. A rigorous opinion therefore needs (a) the front-page References Cited list, (b) the WCCP/discovery-protocol document contents, and (c) the file history — none of which I have.

Explicit uncertainty / what I did not verify

• WCCP draft draft-forster-web-pro-00.txt — cited by the '702 Background; contents not fetched.
• Distributed Director / Local Director / DRP product literature — discussed by the patent's own text; not independently fetched.
• RFC 3036 (LDP) and RFC 3032 (Label Stack Encoding) — referenced within RFC 2547 as references, but the documents themselves were not fetched.
• US 6,330,602 ("Scaleable web server…") — a tool-call limit was hit; not verified.
• US 2005/0149531 A1 (Srivastava et al., "routing data to a load balanced server using MPLS packet labels") — appears to be a same-family publication (priority 2002-03-19); its claims were not retrieved.
• RFC 1546 (Host Anycasting) — not verified this session.
• Dependent claims 13–22' literal text — not retrieved.

If you supply the actual Prior Art section, I'll rebuild the chart reference-by-reference and, if you also provide the file history/References Cited list, I can convert this into a limitation-by-limitation validity opinion with a candid allowance-traversal reconstruction.