US 10841398

I'll research this patent across several sources. Let me run independent searches.

Let me verify the granted claims and check litigation/appellate dockets specifically referencing this patent.

US Patent 10,841,398 — Research Summary

Scope note / disambiguation: This report concerns US 10,841,398 B2 only. My searches also surfaced the similarly-numbered US 10,097,398 B2 ("Point of presence management in request routing," Amazon Technologies) — that is a different, unrelated patent and is excluded here.

---

Bibliographic data

---

Abstract

The full text I retrieved for the grant itself (Google Patents, https://patents.google.com/patent/US10841398/en) was truncated before the Abstract section, so I could not read the abstract of US 10,841,398 verbatim off the grant document. The abstract text used across this patent family (confirmed on the later continuations US 12,284,260 B2 / US 2022/0116477 A1 / US 2025/0310417 A1) reads:

"A content delivery network (CDN) includes a plurality of CDN components including at least one CDN rendezvous mechanism and at least one control core. The CDN components are controlled by control core data from the at least one control core. Some CDN components obtain CDN resources including control core data from at least some other CDN components. The CDN components use the CDN rendezvous mechanism to select one or more CDN components from which to obtain CDN resources."

Uncertainty flag: Because these are the same specification/family, the '398 abstract is very likely identical or near-identical, but I could not confirm it against the '398 grant document directly. Treat the wording above as family-level, not verified-grant-level.

---

Independent claim — plain-language overview

Caveat on claim provenance: I could not open USPTO PatentCenter / the grant PDF itself. The claim set below comes from the Justia page for publication US 2019/0082029, which is cross-identified there as granting as patent 10,841,398 (19 claims). The file history (see below) shows a final rejection followed by an amendment-after-final and a 9/18/2020 Examiner's Amendment before allowance, so the granted claim wording may differ in detail from what I found. Verify against the printed patent before relying on it.

Claim 1 (the only independent claim I could identify; claims 2–19 all depend from it):

A content delivery network (CDN) comprising: a plurality of CDN components including at least one CDN rendezvous mechanism and at least one control core, wherein the plurality of CDN components are controlled by control core data from the at least one control core, wherein at least some particular CDN components are constructed and adapted to obtain CDN resources from at least some other CDN components, and wherein the CDN components use the CDN rendezvous mechanism to select one or more CDN components from which to obtain CDN resources.

In plain language, claim 1 recites four elements acting together:

1. A CDN made of many components, including (a) at least one rendezvous mechanism (in the specification, a DNS-based name server system that steers requests to a "best"/"optimal" location) and (b) at least one control core (the authoritative, distributed configuration database).
2. The CDN components are governed by control-core data — i.e., the control core supplies configuration/control information that the other components obey.
3. CDN elements act as each other's clients and servers — some components are built to pull CDN resources (including control-core data, logs, etc.) from other CDN components, not just from the control core. The specification frames this as treating internal control/config data as ordinary web/cacheable objects with URLs, so the CDN can serve itself ("the CDN is a customer of itself").
4. The CDN's own rendezvous mechanism is reused internally to pick which component to fetch from — i.e., internal CDN-to-CDN retrieval uses the same DNS-based selection machinery used to steer external clients.

Selected dependent claims (context only, not independent): control-core data as distributed resources/objects (claims 4, 8, 14); cache servers that deliver customer resources and are themselves "customers" of the CDN (claims 5–6); distributed control core with distributed consensus (claims 9–11); access to the control core via control-core domain names resolved using the CDN rendezvous mechanism (claims 12–13); control core as the "origin tier" for control objects in a tiered/hierarchical CDN (claim 15); authoritative database of current CDN configuration with voting for consistency (claims 18–19).

---

Priority / family

The '398 is a continuation in a large Level 3 / Sandpiper CDN family. Its own specification states it claims priority through Ser. No. 14/302,865 (filed 2014-06-12, pub. US 2014/0372588 A1), itself a continuation of Ser. No. 13/714,410 (filed 2012-12-14, "Content Delivery Network"), which claimed priority from provisionals 61/570,448 and 61/570,486, both filed 2011-12-14. Continuations issuing from the '398 line include US 11,218,566 B2 (2022-01-04), US 11,838,385 B2 (2023-12-05), US 12,284,260 B2 (2025-04-22), and published application US 2025/0310417 A1 (2025-10-02). (Note: some aggregator records list a priority date of 2011-12-13; the authoritative Google Patents record and specification citations both indicate 2011-12-14.)

---

Litigation and CAFC 2026 dockets

• Litigation flag: The Google Patents record carries a "Family has litigation — first worldwide family litigation filed" flag with a Darts-ip family link (family 48611345). I found a Sandpiper CDN, LLC v. Microsoft Corporation complaint in E.D. Tex., 2:25-cv-00664 (CourtListener docket 70642050), but the excerpts I retrieved discuss other patents in the portfolio (e.g., a "692" patent on tiered servers and a "173" patent on caching); I could not confirm from the retrieved text that 10,841,398 itself is asserted in that case.
• CAFC 2026: I ran targeted searches for a 2026 Federal Circuit docket involving 10,841,398 and for Sandpiper CDN appellate activity. I found no evidence of a 2026 CAFC appeal involving this patent. The CAFC 2026 results returned were unrelated cases (EcoFactor reexam appeal; Nike v. Lululemon). I also found a Level 3 Communications v. Amazon appeal (CAFC No. 23-2181, dismissed by stipulation under FRAP 42(b) on 2024-02-08) — but that involved US 7,945,544 / 7,802,310 / 6,928,442 / 8,099,420, not the '398.

Bottom line: No authoritative source I could reach shows a pending 2026 CAFC proceeding on 10,841,398. This is a negative search result, not proof of absence — my search was not a complete Federal Circuit docket sweep.

---

Prosecution snapshot (from the file-wrapper index)

Application 16/181,010: non-final rejection mailed 2020-02-13; response 2020-06-15; final rejection 2020-06-29; response after final 2020-08-27; Examiner's Amendment / reasons for allowance 2020-09-18; Notice of Allowance data verification completed 2020-09-22; patent granted 2020-11-17. This history is consistent with claim amendments before allowance.

---

Confidence and uncertainty summary

• High confidence: title, patent number, application number, filing date (2018-11-05), issue date (2020-11-17), assignee chain (Level 3 → Sandpiper CDN), inventor list (six inventors), family members, and the subject-matter content (all corroborated by the full Google Patents text plus multiple independent aggregators).
• Moderate confidence: the abstract wording (taken from same-family continuations, not read off the '398 grant) and the exact granted wording of independent claim 1 (taken from a secondary source; an Examiner's Amendment preceded allowance).
• Not established: any 2026 CAFC appeal, and whether 10,841,398 is among the patents asserted in the Sandpiper CDN v. Microsoft case.
• I did not access USPTO PatentCenter or PACER/CAFC dockets directly; I could not complete those direct lookups.

I'll search for litigation involving US patent 10841398 across the sources you mentioned.

Both returned empty. Let me try alternate queries and sources.

Good, that surfaced real leads. Let me dig into the specific cases and confirm which patents are asserted.

I have a step limit warning, but let me run a few final targeted searches to determine whether '398 specifically appears in any of these cases.

I'll verify the structured block's finding with targeted searches before concluding.

Those hits are all different "398" patents (Google 10,134,398; Nuance 9,818,398; RE45,398). Let me confirm with patent-specific searches.

Critical context: Sandpiper CDN is asserting a portfolio of Level 3 CDN patents and Google/Microsoft have filed numerous IPRs — but the patents identified so far are '112, '903, and '573, not '398. Let me pin down whether any IPR targets the '398 patent specifically.

Proceedings overview

Total AIA trial proceedings on file for US 10,841,398: 0. Breakdown: 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution denials. The USPTO Open Data Portal (ODP) returns no AIA trial proceedings for this patent, and independent web searching surfaced no IPR, PGR, or CBM petition naming the '398 patent. For a defendant today, that means every claim of the '398 patent remains presumption-of-validity-intact, untested at the PTAB, and un-narrowed by any adverse judgment — you get no free cancellation to point to, and you inherit no answer brief or FWD to build on. Equally, you face no § 315(e)(2) estoppel and no § 315(b) time bar attributable to any '398 proceeding, so the full prior-art field is open to you.

⚠️ Contradiction / clarification flag: The Google Patents family page for '398 carries a "Family has litigation" flag pointing to Darts-IP family 48611345. That is litigation, not PTAB activity, and it does not by itself put '398 at the Board. Treat it as a monitoring trigger, not as evidence of an AIA proceeding.

No per-proceeding entries

There is nothing to list. I will not manufacture proceeding numbers, panels, or dispositions. Note the false-positive trap: searches for "398" and "PTAB" return three different patents that are frequently confused with this one:

• U.S. 10,134,398 (Google, hotword detection) — IPR2023-00118 / IPR2023-00119, Sonos v. Google; FWDs held claims unpatentable, reversed and remanded by the Federal Circuit in Google LLC v. Sonos, Inc., Nos. 2024-2119, 2024-2120 (Fed. Cir. 2026-06-09), 24-2119. https://law.justia.com/cases/federal/appellate-courts/cafc/24-2119/24-2119-2026-06-09.html
• U.S. 9,818,398 (Nuance, ASR error flagging) — Nuance Preliminary Response in an IPR filed 2018-11-15.
• RE45,398 (welding/monitoring) — unrelated IPR.

None of these is the Level 3 / Sandpiper content-delivery patent. Do not cite them as '398 precedent.

Strategic summary

Claim status: UNTESTED — all claims. No claim of US 10,841,398 has been canceled, confirmed, or construed by the PTAB in any AIA trial. Practically, that cuts both ways. You cannot argue in a demand-letter response that "the PTAB already killed the claims you're citing" — because it did not. Conversely, Patent Owner (Sandpiper CDN, LLC, which acquired the portfolio from Level 3 Communications, LLC in April 2024) has no FWD-backed claim construction to hide behind, and no statutory estoppel has been created.

Context that matters: the owner is running an active campaign against this exact portfolio — just not on this patent (yet). Sandpiper's 2024–2025 campaign asserts Level 3 CDN patents against Google (C.D. Cal., No. 2:24-cv-03951, filed 2024-05-10), Comcast (E.D. Tex., No. 2:24-cv-00886, filed 2024-11-01), and Microsoft (E.D. Tex., No. 2:25-cv-00664, filed 2025-06-26). Defendants have responded with a large IPR wave — Google filed IPR2025-00806, -00826, -00846, -00852(‑00952), -00860, -00969, -01010; Microsoft filed IPR2026-00095, -00174, -00180, -00190. Every patent number I could positively identify in those proceedings is a sibling/portfolio patent, not '398: U.S. 9,021,112, U.S. 8,478,903, U.S. 10,924,573, U.S. 8,719,886, and U.S. 8,595,778 / 8,719,886 (held § 101-invalid in the district court). See https://www.docketalarm.com/cases/PTAB/IPR2026-00095/Microsoft_Corporation_v._Sandpiper_CDN_LLC/ and https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00826. I could not verify the complete asserted-patent list, and I did not confirm that '398 appears in any complaint. That is the single most important open question below.

Estoppel landscape: wide open. Because no IPR has been instituted on '398, § 315(e)(2) estoppel attaches to nobody with respect to this patent. All § 102/§ 103 grounds, all art, and all combinations remain available, including art already used against the sibling patents (e.g., Dilley, Kenner, Glines, Newton‑471, Mowry declaration testimony) — subject only to § 325(d) discretion and to the fact that Patent Owner may argue those references were already considered in some Office proceeding on the related family. One timing caveat: if '398 was named in the Google complaint served 2024-05-10 or the Comcast complaint served 2024-11-01, then a § 315(b) one-year bar may already have run for those parties — verify against the complaints before assuming a petition is timely.

Pattern signals. (1) No defensive aggregator (Unified Patents, RPX, AST) appears in the chain for '398; Unified's portal surfaces '398 only as a patent profile page, which is not evidence of an IPR filing. (2) No serial-petition pattern on '398 because there is no first petition. (3) The portfolio-level pattern is a classic NPE acquisition (Delaware entity formed March 2024, assignments from Level 3 in April 2024) followed by a multi-defendant assertion campaign and a well-funded IPR counterattack. (4) The '398 patent has live continuations/divisional family members — US20190082029A1, US 11,218,566, US 11,838,385, US 12,284,260, and US 2025/0310417 A1 — which can carry overlapping claims into new suits even if a parent is invalidated. Monitor those too.

Recommended next steps

1. Verify the asserted-patent list directly. Pull the three complaints — Sandpiper CDN, LLC v. Google LLC, No. 2:24-cv-03951 (C.D. Cal., 2024-05-10) (first amended complaint, Dkt. 45-1); Sandpiper CDN, LLC v. Comcast Cable Commc'ns LLC, No. 2:24-cv-00886 (E.D. Tex., 2024-11-01); Sandpiper CDN, LLC v. Microsoft Corp., No. 2:25-cv-00664 (E.D. Tex., 2025-06-26) — and confirm on the face of each whether US 10,841,398 is pleaded. If it is, compute your § 315(b) deadline immediately (one year from service of the earliest complaint alleging infringement of that patent by you, a real party in interest, or a privy).

2. Run the patent-specific PTAB check yourself, at the source. Search the patent number and application number 16/181,010 at PTAB E2E (https://ptab.uspto.gov) and in the PTAB Decisions database (https://developer.uspto.gov/ptab-api or the PTAB public search). I could not exhaustively query the full PTAB docket for this number due to a search-step limit; the ODP structured source is authoritative and says zero, but a manual E2E check takes two minutes and removes all doubt.

3. If you are a defendant and you need a cancellation-based defense: there is none to rely on. No FWD exists for '398, so you cannot quote a disposition canceling claims 1–5 (or any claims). Your options are (a) file first and win institution, or (b) defend in district court on invalidity, and note that the same PTAB panels (APJs Weatherly, McShane, Cygan; Tech Center 2400 / Art Unit 2451–2457) are working through the sibling patents, with Google's institution decisions landing 2025-11-07 through 2025-12-01 and Director Review of those denials refused 2026-01-12 (see IPR2025-00806/-00826/-00969/-01010, Paper 20).

4. Do not over-read the zero. The absence of PTAB activity on '398 is a negative signal, not a comforting one: it most likely means the patent has not yet been asserted against a defendant willing to fund an IPR, or has been held in reserve. Well-asserted CDN patents in this portfolio have attracted IPRs within roughly 12 months of complaint service. If Sandpiper amends a complaint to add '398, expect a petition cycle to follow — and plan your prior-art search now rather than after the year runs.

Bottom line: US 10,841,398 has no AIA trial history whatsoever — no institution decisions, no FWDs, no settlements, no appeals. Asserted against you today, it is fully intact and fully attackable, with no estoppel constraining either side and no statutory clock running on it yet. The controlling caveat is procedural, not substantive: confirm from the complaints whether you have already been served on this patent, because the § 315(b) clock — not the merits — is what will decide whether you can reach the Board at all.

I'll research the ownership chain for US 10841398 using the patent text I have plus targeted searches.

Strong leads. Let me pin down the assignment reel/frame details, correspondent of record, and Sandpiper CDN's corporate footprint.

Found the key reel/frame. Let me get the correspondent of record and the earlier inventor-assignment details.

Ownership and Assignment Record — US 10,841,398 B2

Bottom line up front: '398 is a Level 3 Communications continuation in the Sandpiper CDN family. The chain has exactly two recorded links I can verify: (1) inventors → Level 3 Communications, LLC, and (2) Level 3 Communications, LLC → Sandpiper CDN, LLC. The second is a clean fire-sale transfer from an operating CDN operator to a newly formed licensing vehicle that began suing within weeks. This is an NPE chain, and the strongest evidence is not the LLC's name — it is the formation date, the licensing-only self-description, and the pre-litigation timing.

---

Inventors

All six appear as assignors in the 2018-11-14 recorded assignment to Level 3 (see below), confirming they were under obligation to assign to Level 3. The named inventors are consistent across the continuation family (US 11,218,566; US 11,838,385; US 12,284,260; US 10,187,491, etc.).

Unusual patterns: I cannot confirm, with high confidence, inventor departure dates from Level 3/Lumen, so I am not asserting the "all inventors depart within 12 months" pattern. What I can say is that the inventors are a stable team that appears on many sibling filings, which is typical of an operating-company R&D group rather than a hired-inventor scheme. No finding on the departure signal.

Note on provenance: the '398 patent identifies the Sandpiper Networks figureheads (David Farber, Andrew Swart) as inventors only on sibling patent US 8,478,903, not on '398. Do not conflate the 1990s Sandpiper origin story with the inventorship of this 2011-priority filing.

---

Original assignee

Level 3 Communications, LLC (Broomfield, CO). Per the issued patent, original assignee of record.

• Primary line of business: Tier-1 global network/telecom operator; operated a large commercial CDN that it built substantially on the acquired Sandpiper Networks / Digital Island / Savvis CDN assets and IP (Level 3 acquired the Savvis CDN business and its patents in Dec 2006 for ~$132.5M cash).
• Ship product embodying the claims? Yes — Level 3 ran a production CDN for years, and its CDN platform (later branded under Lumen) served major streaming customers. The '398 claims recite control/framework mechanisms used in that platform.
• Current status: Operating, but no longer in the CDN business. Level 3 was acquired by CenturyLink in 2017; the combined company rebranded as Lumen Technologies in 2020. Lumen exited the CDN market in 2023 and sold CDN assets/IP. So the original assignee persists as a going concern but has divested the relevant line of business — the classic setup for a portfolio disposal.

---

Assignment timeline

Two recorded conveyances are documentable for this family; I recovered the reel/frame for the NPE link but not the reel/frame for the inventor-to-Level-3 link, and not the correspondent of record (see caveats).

• Executed ~2011-12-14 / recorded 2018-11-14 — Reel not recovered (Assignment of interest)

  • Conveyance: Assignment (Assignment of Assignors' Interest)
  • Assignor: Laurence R. Lipstone; William Crowder; Jeffrey G. Koller; David Fullagar; Maksim Yevmenkin; Christopher Newton
  • Assignee: LEVEL 3 COMMUNICATIONS, LLC
  • Correspondent: Not recovered from available sources. (USPTO Assignment Center lists it; I could not retrieve the cover-sheet attorney/firm text.)
  • Context: Ordinary employment/obligation-to-assign — original corporate ownership of a continuation application. The 2018 recording date reflects the 2018-11-05 filing of this continuation; the priority invention dates to 2011-12-14.

• Executed 2024 (pleaded 2024-03-29; RPX reports the assignment instrument dated 2024-04-24) / recorded 2024-04-26 (RPX) — Reel 068256 / Frames 0091–0115

  • Conveyance: Assignment
  • Assignor: LEVEL 3 COMMUNICATIONS, LLC
  • Assignee: SANDPIPER CDN, LLC
  • Correspondent: Not recovered / not confirmed. On the analogous Level 3 → Optic153 transfers, the recorded signatory was John T. Meli Jr. (Equitable IP), but I could not confirm an analogous signatory/correspondent for the Sandpiper reel. Do not treat the Equitable IP correspondent as applying to the Sandpiper chain — that is a different Level 3 NPE transfer.
  • Context: Fire-sale / portfolio disposal at market exit. Level 3 (Lumen) exited the CDN business in 2023 and moved 80+ US patents (RPX count) to Sandpiper CDN in April 2024. This reel/frame is corroborated by an actual court/PTAB exhibit: Microsoft's IPR petitioner exhibit EX1036 — "Patent Assignment Cover Sheet from Level 3 Communications, LLC to Sandpiper CDN, LLC, Reel No. 068256, Frames 0091-115."

Discrepancy to flag (per your instruction): Google Patents' legal-events feed lists the Sandpiper acquisition as "2024-07-09 Assigned to SANDPIPER CDN, LLC." This does not match the RPX reporting (dated 2024-04-24 / recorded 2024-04-26) or the pleaded sale date (2024-03-29). Possible explanations: a separate/corrective recordation, a second tranche of the assignment, or Google's feed reflecting a later record event. The existence of the transfer is not in doubt (Reel 068256/0091-115 is a live record cited in an IPR), but I cannot reconcile the exact date across sources. Treat the 2024-04-26 recordation as the best-supported date.

Conveyance types not present: No Security Agreement, Merger, Change of Name, License, or Release recordings recovered for this chain.

---

Timeline diagram

timeline
    title Ownership of US 10841398
    2011 : Priority application invented
         : Original assignee Level 3
    2018 : Continuation filed
         : Inventors assign to Level 3
    2020 : Patent issued to Level 3
    2024 : Sandpiper CDN LLC formed in Delaware
         : Level 3 to Sandpiper CDN assignment
         : First suit filed against Google
    2025 : Suits against Comcast and Microsoft
         : Microsoft IPRs filed

---

NPE / troll-pattern signals

---

Verdict

NPE — high confidence (2+ strong signals).

Two independent strong signals are present. (1) Shell-entity transfer: an operating CDN operator (Level 3/Lumen, ~1,282 patents in the class) moved 80+ assets to Sandpiper CDN, LLC, a Delaware LLC formed 2024-03-21, whose own website markets "licensing opportunities" and staffs an "IP Licensing" manager — Reel 068256/0091–0115. (2) Pre-litigation transfer: the assignment (executed 2024-03-29, recorded 2024-04-26) landed roughly 1–2 months before the first infringement complaint (Google, May 2024), with Comcast (2024) and Microsoft (2025) suits following. The confirming reel/frame is not merely a registry entry — it is cited in litigation as IPR petitioner exhibit EX1036, which raises confidence well above a naming-based guess.

Caveats I am flagging rather than papering over: (a) I could not recover the correspondent of record on either link, so signal 3 is unresolved; (b) Google Patents shows a conflicting assignment date of 2024-07-09 versus the 2024-04-26 recordation reported by RPX; (c) Sandpiper is a newly active plaintiff, so signal 2 rests on RPX's "high-frequency plaintiff" tracking rather than a match to the legacy NPE list.

Verify at: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search patent number 10841398) — and the Family-recorded cover sheet at the PTAB exhibit EX1036 in IPR2026-00190 (Microsoft Corp. v. Sandpiper CDN, LLC).

I'll research the specific patent and its cited prior art. Let me run targeted searches.

Let me search for the specific patent citations (prior-art references) listed on the '398 patent.

Let me retrieve the full cited-references list from FreePatentsOnline and the USPTO record.

Prior-Art Analysis — US 10,841,398 B2

(Building on the "Patent summary" and "Litigation summary" sections already produced. Those sections are not repeated; where this analysis depends on them, I cross-reference. One contradiction is flagged in §6.)

---

1. Confirmation of the searched record

The search returns a single match for US 10,841,398 B2 and nothing for adjacent/similar numbers as requested:

Do-not-confuse note (per the strict literal-interpretation rule): the search corpus repeatedly intermixes three distinct identifiers that must not be collapsed — US 10,841,398 (this patent), US 10,097,398 (Amazon, "Point of presence management in request routing" — already flagged as unrelated in the prior section), and the family "…4xxx" numbers 10,841,177 / 10,841,400 (Level 3, different patents). References below are keyed to 10,841,398 only.

---

2. The cited references I was able to verify

Sourcing caveat (important for § 102 reliance): The full Google Patents text I retrieved was truncated before the "Patent Citations" block, and I was not able to open USPTO PatentCenter's "References Cited" pane or the grant PDF. The only grant-level list I could retrieve came from the third-party aggregator uspto.report (https://uspto.report/patent/grant/10,841,398), which reproduced the "References Cited" table. I retrieved the first ~38 entries of that table before the page output ended. I therefore cannot certify that this is the complete list, nor which entries were cited by the examiner vs. by the applicant, nor which were actually applied in the 2020-02-13 non-final and 2020-06-29 final rejections. Treat everything below as potentially cited prior art pending confirmation against the file wrapper.

References retrieved from the '398 "References Cited" table:

A separate, family-level citation set. The Google Patents "Patent Citations (220)" block for the sibling US 9,456,053 B2 (same specification, same priority) surfaced the following references — these are material background because the '398 shares the '053 specification, but they are the '053's citation list, not confirmed as the '398's:

• US 8,060,613 B2 / US 2008/0215735 A1 — Level 3, "Resource invalidation in a content delivery network" (1998 priority)
• US 6,654,807 B2 — Cable & Wireless, "Internet content delivery network"
• US 7,054,935 B2 — Savvis, "Internet content delivery network"
• US 7,945,693 B2 / US 7,949,779 B2 — Level 3, "Controlling subscriber information rates in a CDN"
• US 7,523,181 B2 — Akamai, "Method for determining metrics of a content delivery and global traffic management network"
• US 7,860,964 B2 — Level 3, "Policy-based content delivery network selection"
• US 7,010,578 B1 — Akamai (2000-09-21)
• US 2007/0271385 A1 — Akamai, "Managing web tier session state objects in a CDN"

---

3. Most relevant prior art, mapped to the '398 claims

§ 102 framework applied. Anticipation under 35 U.S.C. § 102 requires a single reference disclosing every element of a claim, arranged as claimed. Because claim 1 is a markedly broad apparatus claim ("a plurality of CDN components including at least one CDN rendezvous mechanism and at least one control core… particular CDN components… obtain CDN resources from… other CDN components… use the CDN rendezvous mechanism to select one or more CDN components from which to obtain CDN resources"), the honest answer is that the most dangerous art is the applicant's own incorporated-by-reference CDN patents, not the miscellaneous 2002–2005 applications in the examiner's list.

3.1 Closest art — the applicant's own incorporated prior art (spec ¶¶[0003]–[0014])

These are named in the '398 specification's "Incorporation by Reference" list and expressly relied on to describe the rendezvous, load-balancing and CDN-selection mechanisms that claim 1 recites. They are not shown as examiner citations in the table I retrieved, but they are the most technically on-point:

§ 102 note: Because these are incorporated by reference into the '398's own specification, they are not "prior art against" the '398 in the ordinary sense — they are the applicant's admission of what was already known. Their practical significance is that they bound how broadly claim 1 can be read, and they are the references most likely to appear in an IPR petitioner's obviousness combination.

3.2 CDN-specific cited publications

• US 2002/0010798 A1 — Ben-Shaul et al., pub. 2002-01-24, "Differentiated content and application delivery via internet." Describes distributed content/application delivery across intermediaries. Potentially anticipates/relevant to claim 1 (plurality of delivery components obtaining content from other components) and claims 3–4, 7–8 (control data as distributed resources) if it discloses centralized control data governing the delivery components. This is the single CDN-overlay application in the retrieved list with confirmed subject matter.
• US 2002/0091801 A1 — Lewin, pub. 2002-07-11, "Extending an internet content delivery network into an enterprise." Enterprise extension of a CDN; bears on claim 1 (component-to-component resource acquisition) at the CDN boundary.
• US 2004/0073596 A1 — Kloninger et al., pub. 2004-04-22. Akamai-lineage CDN subject matter; potentially relevant to claim 1 and claims 17–19 (configuration/maintenance). Title not verified from retrieved data.
• US 2004/0162871 A1 — Pabla et al., pub. 2004-08-19. Distributed-caching / peer-distribution subject matter; potentially relevant to claims 14–17 (control objects obtained via the CDN itself). Title not verified.

3.3 Distributed-consensus / voting art (claims 9–11, 18–19)

The '398's dependent claims 9–11 (distributed control core using a distributed consensus algorithm) and 18–19 (authoritative database maintained with voting) map directly onto the consensus references the specification itself cites in ¶¶[0165]–[0167]:

• US 2003/0065708 A1 — Jacobs, pub. 2003-04-03 (likely the pre-grant publication of the Jacobs et al. distributed-consensus work cited in spec ¶[0165]). Bears on claim 11 and claim 19.
• Paxos-family patents named in spec ¶[0167]: US 7,856,502; US 7,797,457; US 7,711,825; US 7,698,465; US 7,620,680; US 7,565,433; US 7,558,883; US 7,555,516; US 7,249,280; US 6,463,532. These are incorporated "for the purpose of describing the Paxos algorithm," so they are material to claims 9–11 and 18–19.

§ 102 caution: a Paxos reference alone does not anticipate claim 1, because it lacks the CDN-component and rendezvous elements; its § 102/§ 103 exposure is confined to the consensus dependent claims.

3.4 General networking/publication art (low individual relevance)

Entries #6–14, #16–38 in the table are 2002–2005 published applications in the general distributed-computing, web-services, peer-to-peer and caching space. On the retrieved data I cannot responsibly assign each a specific anticipating claim, because I have neither their titles nor their disclosures, and § 102 requires element-by-element disclosure. They are best characterized as § 103 obviousness fodder for secondary features (session objects, sequence/handler processing, peer caches) rather than § 102 anticipators of claim 1.

---

4. Which claims the cited art could plausibly reach

---

5. Bottom line on the most relevant prior art

1. The most relevant prior art is the applicant's own incorporated-by-reference CDN portfolio — principally US 6,185,598; US 6,654,807; US 7,054,935; US 7,822,871; US 7,860,964 (rendezvous/selection) and US 8,015,298 / US 8,489,750 (cluster architecture). These map onto claim 1's rendezvous and component elements better than any of the third-party 2002–2005 applications the examiner cited.
2. The best third-party candidate in the Retrieved citation table is US 2002/0010798 A1 (Ben-Shaul et al.), with US 2004/0073596 A1 (Kloninger) and US 2004/0162871 A1 (Pabla) as secondary candidates.
3. The consensus/voting dependent claims (9–11, 18–19) are covered by US 2003/0065708 (Jacobs) and the Paxos-family patents named in the specification.
4. No single retrieved reference appears to anticipate claim 1 in full, chiefly because none is confirmed to disclose the internal reuse of the CDN's own rendezvous mechanism to select the source CDN component — the apparent point of novelty. Instead, the realistic § 103 attack is a combination: a rendezvous/CDN-selection reference + a centralized-control-distribution reference + a component-to-component-fetch reference.

---

6. Confidence and uncertainty (and one contradiction)

• High confidence: the identity of US 10,841,398 B2; the existence and identity of the incorporated-by-reference list in spec ¶¶[0003]–[0014]; the Paxos/consensus references in ¶¶[0165]–[0167]; the identity of the first ~38 entries of the examiner/applicant "References Cited" table.
• Moderate confidence: the titles/subject matter of Ben-Shaul and Lewin (confirmed via the family citation list); the identification of US 2003/0065708 (Jacobs) with the Jacobs distributed-consensus work cited in ¶[0165] (chronologically consistent, not confirmed).
• Not established: the complete citation list for the '398; the split between examiner-cited and applicant-cited references; which references were actually applied in the 2020-02-13 non-final or 2020-06-29 final rejections; and the specific claim(s) each reference was used against in prosecution. I could not open PatentCenter / the grant PDF.
• Contradiction flagged: as noted in the prior "Patent summary," the priority date is recorded as 2011-12-14 on Google Patents and in the specification, but the Unified Patents portal snippet displayed 2011-12-13 for US-10841398. Per the operating rules I do not auto-correct; I note the discrepancy and treat 2011-12-14 as the specification-supported date while marking the 12-13 variant as an unresolved aggregator discrepancy. This matters to § 102 because a one-day shift can move the critical date relative to a close-in-time reference (e.g., the applicant's own US 2014/0372588-family publications).

Grounding URLs:

• Grant: https://patents.google.com/patent/US10841398B2/en
• Retrieved citation table: https://uspto.report/patent/grant/10,841,398
• Family citation list (220): https://patents.google.com/patent/[US9456053B2](/patent/US9456053B2)/de
• Spec priority/incorporation statements: US 2021/0067601 A1 (app. 17/095,325) — https://uspto.report/patent/app/20210067601

I'll verify a few key reference identities and check for any obviousness-specific treatment of this patent family, then build the § 103 analysis on the prior-art section already produced.

Obviousness Analysis — US 10,841,398 B2 under 35 U.S.C. § 103

Builds on the Prior-Art, Litigation, and PTAB sections already produced. Prior-art reference identities below are those established in the Prior-Art section unless this analysis newly verifies them; newly verified facts are cited to their sources. Contradictions found against earlier sections are flagged in §11.

---

0. Threshold issues that change the § 103 analysis

0.1 Which § 102/§ 103 regime applies. This matters because pre-AIA § 103(c) contains a common-ownership carve-out that the AIA does not (for § 102(a)(1) art).

• Application 16/181,010 was filed 2018-11-05, but every claim is a continuation claim entitled to 2011-12-14. Under AIA § 3(n)(1), the first-inventor-to-file provisions reach an application only if it "contains or contained at any time" a claim with an effective filing date on or after 2013-03-16. If all claims here are supported by the 2011-12-14 priority chain (Ser. 14/302,865 ← 13/714,410 ← provisionals 61/570,448 / 61/570,486), pre-AIA § 102/§ 103 govern. I flag this as moderate confidence — I did not read the '865 file wrapper to confirm no post-2013 new matter was ever claimed.
• Practical consequence: pre-AIA § 103(c)(1) blocks obviousness rejections built on § 102(e)/(f)/(g) art that was commonly owned at the time of invention. The applicant's own family art (Farber '598, ATC '871, Lewin-family) is therefore only safe to use if it qualifies under § 102(a)/(b). It does: see §0.2.

0.2 Critical dates for the principal references. For a 2011-12-14 priority date, the pre-AIA § 102(b) one-year bar runs from 2010-12-14:

The ATC '871 date is the one to police: it issues 50 days inside the § 102(b) window. If the critical date were the aggregator-variant 2011-12-13 flagged in the Prior-Art section, the analysis is unchanged (still outside the one-year bar). If the priority chain were ever broken back to a 2012 date, '871 remains § 102(b) art regardless.

0.3 § 311(b) constraint if this is an IPR. Grounds are limited to § 102/§ 103 "only on the basis of prior art consisting of patents or printed publications." System/prior-use art (e.g., Level 3's actual production CDN, or the real-world Akamai/Speedera deployments) is not available. Every ground below is built from patents and printed publications for that reason.

0.4 Claim-set provenance caveat (carried forward, not repeated). The Prior-Art section already flags that the granted claim wording is reconstructed from a secondary source and that a 9/18/2020 Examiner's Amendment preceded allowance. Every limitation mapping below is therefore provisional pending a check against the printed patent.

0.5 Claim construction that governs the scope. Under Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc), the specification is the single best guide:

• "CDN resources" — the '398 specification expressly defines its universe of objects as "[e]ach object … treated as a web object or resource, with, e.g. the control core acting as the 'origin tier' for such objects" and "[e]ach CDN object has a URL … can be requested, filled, invalidated, refreshed, etc." (spec, "Executable Resources"/"Control core" passages, https://patents.google.com/patent/[US10841398B2](/patent/US10841398B2)/en). Under the broadest reasonable reading, "CDN resources" covers any object the CDN serves internally — configuration, scripts, logs, invalidation journals, and ordinary customer content pulled from a peer/parent.
• "constructed and adapted to" — construed as "configured to"/"capable of" (In re Giannelli, 739 F.3d 1375, 1378 (Fed. Cir. 2014)). This is a capability recitation, not a structural one; it does not require that the component actually perform the fetch.
• "control core" / "control core data" — the spec's own text: "the control core 108 keeps the authoritative database of the current CDN configuration"; caches obtain GCO and CCSs from it via HTTPS pull.
• "rendezvous mechanism" — the spec defines it functionally and broadly: "the rendezvous system 104 … is preferably implemented using the DNS and comprises one or more DNS name servers", and the "best"/"optimal" location "may even be another CDN or network or a server outside the CDN 100."

This construction is adverse to the Patent Owner: read this broadly, claim 1 is close to a general statement of how a CDN works.

---

1. Level of ordinary skill in the art (Graham factor 3)

A POSITA at the 2011-12-14 critical date would hold a B.S. in CS/EE (or equivalent) plus 3–5 years of experience designing or operating distributed content-delivery, caching, or large-scale web-serving infrastructure, or a master's degree with 1–2 years. The person would be familiar with: DNS-based request routing and load balancing; HTTP/1.0–1.1 and HTTPS; hierarchical/peer cache fill; reverse-proxy and content-router design; and distributed-replica consistency (quorum/voting, leader election, two-phase commit).

This is not a high-skill bar, and that cuts against the Patent Owner. The '398's claims do not recite any algorithm, protocol, data structure, API, latency bound, or measured result — only an architectural arrangement of functions. A POSITA here is "a person of ordinary creativity, not an automaton." KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398, 421 (2007).

---

2. Claim 1 decomposed into limitations

The only genuinely contestable limitation is L4 — specifically, the reuse of the externally facing rendezvous mechanism to select an internal source. L1–L3 are each individually old; L4 is the point the examiner appears to have relied on for allowance (consistent with the 9/18/2020 Examiner's Amendment after a final rejection). The whole § 103 case therefore lives or dies on whether the art renders L4 obvious — and on whether the specification's own admissions supply the missing motivation.

---

3. The specification supplies most of its own motivation

Three passages in the '398 specification are, in effect, admitted prior art and admitted design rationale. They are quoted here because they are the strongest § 103 motivation evidence in the record (usable in an IPR as evidence of what a POSITA knew and why the combination would be made; the challenged spec is in evidence as Ex. 1001):

1. L4 itself is described as an available design choice: "the selection of the 'best' or 'optimal' control core component to serve the cache server 1608 may be made with the same rendezvous and selection mechanism(s) used to direct client requests to servers in the CDN." The patent frames this as the same mechanism — not a new one.
2. The control core as an internal origin/destination resolved by DNS: "the domain name control.fp.net will be used for the control core 108 … the DNS server(s) will provide the client with a rendezvous to a 'nearby' control core server or servers."
3. Component-to-component control-plane delivery, expressly: "resources that originate at the control core may be served to a cache by a peer or parent rather than requiring that each cache pulls directly from the control core," and "the CDN may be considered to be a customer of itself."

A claim that recites, as its point of novelty, an arrangement the specification describes as conventional and desirable is on very weak obviousness ground. In re Nomiya, 509 F.2d 566, 571 (CCPA 1975) (applicant's own specification statement of the prior art is admissible against him).

---

4. Grounds of rejection

All grounds are framed as IPR-style § 103 grounds (patents/printed publications only).

Ground 1 (primary, strongest) — Farber (US 6,185,598) + ATC (US 7,822,871) + Ben-Shaul (US 2002/0010798 A1)

Newly verified identity: US 6,185,598 B1, "Optimized network resource location," inventors Farber, Greer, Swart, Balter, assignee Digital Island, Inc., filed 1998-02-10, issued 2001-02-06. (https://uspto.report/patent/grant/[6185598](/patent/6185598); https://patents.google.com/patent/JP2013514744A/en). Its continuation line runs through Ser. 11/065,412 → US 8,478,903. This is not a peripheral reference — Farber and Swart are Sandpiper Networks' founders, the very predecessors Sandpiper CDN, LLC invokes in its complaints (https://litigation.rpxcorp.com/news/80905-it-s-not-every-patent-complaint-that-name-checks-the-starr-report-and-big-bad-voodoo-daddy).

US 7,822,871 B2, "Configurable adaptive global traffic control and management," is the ATC patent the '398 specification names as its own rendezvous mechanism (spec: "An exemplary rendezvous system 104 is described in … U.S. Pat. No. 7,822,871"). Its claims recite an ATC administrative framework comprising … "a network operation center", an AMA policy-management mechanism, an ATC name server network of domain name servers that return IP/CNAME "based on the geographic location of the client and at least one policy," broadcast of policies to the name servers, dynamic polling of load-share/load-shed policies from managed servers, and monitoring mechanisms that send events to the framework (https://insight.rpxcorp.com/patent/[US7822871B2](/patent/US7822871B2)).

US 2002/0010798 A1, Ben-Shaul et al., "Differentiated content and application delivery via internet" (granted as US 6,976,090 B2, Actona Technologies), teaches an edge server disposed in an external DNS zone that "download[s] said directives of said service policy from said policy control server via said data network" and derives its served content "according to directives of a service policy that resides at said policy control server," plus an edge DNS server configured as authoritative that cooperates with a regional DNS server to resolve a user request to the edge server (https://patentimages.storage.googleapis.com/75/89/fa/aab384b72fddd3/US20020010798A1.pdf; https://webapp1.dlib.indiana.edu/virtual_disk_library/index.cgi/[5628977](/patent/5628977)/FID979/OG/html/1301-2/us06976090-20051213.html).

Mapping:

Why L4 is met. Farber's reflector is a CDN-side component that (i) needs data (tables) to function, (ii) obtains that data from another CDN component (its contact repeater), and (iii) does so by resolving a domain name representing the repeater network via DNS — the same DNS infrastructure that steers clients. '871 supplies the centralized, policy-driven control plane; Ben-Shaul supplies the express teaching of edge components downloading policy directives from a control server as URL-addressable objects.

Motivation (MPEP § 2143 rationales tagged):

• (C)/(D) — known technique, same way, ready for improvement. Farber already distributes control-plane tables (load, link-cost, subscriber) over the same fabric that serves content, selected via DNS. Applying that identical mechanism to the control core's configuration objects (GCO, CCSs) is the paradigm case: "if a technique has been used to improve one device, and a person of ordinary skill in the art would recognize that it would improve similar devices in the same way, using the technique is obvious." KSR, 550 U.S. at 417.
• (F) — design incentives / market forces. A CDN's control plane must reach thousands of caches; the reference architecture already scales the data plane this way. Reusing it avoids a second, bespoke distribution network.
• Applicant's own admission. The spec states L4 is performed "with the same rendezvous and selection mechanism(s) used to direct client requests to servers in the CDN." That is a motivation, not a hindsight reconstruction.
• (G) — predictable variation. No new protocol, packet format, or algorithm is claimed; the result (fetch a config object from a DNS-selected peer) is exactly what the combined references already do for content and tables.

Reasonable expectation of success. Farber's system, '871's system, and Ben-Shaul's system were all commercially deployed in the same field at the same time. Nothing in the claim presents a technical hurdle.

---

Ground 2 (family-tested template) — Farber '598 + Ben-Shaul '798 + Lewin (US 7,010,578 B2 / US 2002/0091801 A1) + Gupta (US 6,405,252)

This ground mirrors the combination Microsoft actually pleaded against the familial sibling US 9,456,053 ("Configuring Content Delivery Networks") in IPR2026-00190: "Petitioner argued that Lewin disclosed the foundational elements … including a CDN with a control core (a Network Operations Center and registration server) and a plurality of cache servers serving content for customers," with Wein supplying global/customer metadata configuration (https://ai-lab-cl-prod.azurewebsites.net/case/ptab/IPR2026-00190/doc/summary/1).

Why this matters here: US 9,456,053 is a 2011-12-14 priority family member sharing the '398 specification (the Prior-Art section already noted the shared citation list, e.g. https://patents.google.com/patent/[US9456053B2](/patent/US9456053B2)/de; https://www.freepatentsonline.com/[9451045](/patent/9451045).html). Microsoft's alternative Ground 2 there — Liu (US 2010/0257258) + Devanneaux (US 2007/0156845) — maps global configuration data from a central authority plus granular per-customer cache configuration. Substituting Lewin (which the Prior-Art section already identified as a cited reference on the '398 itself, US 2002/0091801 A1, "Extending an internet content delivery network into an enterprise") into the '398 analysis gives a control core + cache servers + third-party components joining the CDN — with Gupta supplying "a network of Point of Presence (POP) servers, each with a name server and multiple content servers (WebCache servers)," as argued in IPR2025-00826 (https://ai-lab.exparte.com/case/ptab/IPR2025-00826/doc/1004).

Motivation. (A) Combining prior-art elements by known methods to yield predictable results: Lewin adds components to an existing CDN; the control-distribution mechanism necessarily extends to them. (E)/(F) — the competitive/market pressure to expand CDN reach and serve more customers from the same control plane.

---

Ground 3 (rendezvous-only, minimalistic) — Glines (WO 2001/039470) + Gupta (US 6,405,252) + Ben-Shaul '798

The combination the Board instituted in IPR2025-00826 against US 9,021,112 ("Content Request Routing and Load Balancing for Content Distribution Networks"). Per the institution record: "Glines taught … multiple DNS servers sharing a common anycast address … respond to a hostname query with the IP address of a peered content server," and Gupta supplied "a network of Point of Presence (POP) servers, each with a name server and multiple content servers … and … using an ISP's client DNS server to resolve hostnames." Petitioner's motivation: "to improve content delivery performance … enhance scalability, reliability, and load distribution"; success expected because it involved "applying known networking principles (e.g., server replication, ISP DNS resolution) to analogous systems." Chung (US 6,470,389) supplied cluster definitions.

Application to '398: Glines/Gupta supply L1 and the rendezvous function; Ben-Shaul supplies L3/L4 (edge components fetching policy-directive objects from a control server, via a DNS-based redirection). Motivation: identical to Ground 1 — reuse of the routing layer for internal fetches.

---

Ground 4 (tiered-CDN variants; claims 14–17) — Dilley + Pai + Wang (as instituted in IPR2025-00860) + Farber/Ben-Shaul

Per the Nov 25, 2025 Institution Decision in IPR2025-00860 (US 10,924,573), the Board credited the Petitioner's showing that "Dilley discloses a content delivery network ('CDN') using a tiered distribution service arranged in a hierarchy structure" and that "the edge server will request the content from other servers, that is, peer servers, parent servers, or the origin server" — the exact peer/parent/origin fill logic that the '398's control core occupies as "origin tier" for control objects. Pai adds dynamic popularity-driven redistribution; Wang adds a three-level escalation hierarchy. (https://www.docketalarm.com/cases/PTAB/IPR2026-00095/Microsoft_Corporation_v._Sandpiper_CDN_LLC/02-13-2026-Petitioner/Exhibit-1034-.../; I could not verify the Dilley/Pai/Wang patent numbers from the retrieved excerpt — cite them as Ex. 1003/1004/1005 of IPR2025-00860.)

Motivation for claims 14–15: treating the control core as the "origin tier" for control objects is a straight substitution of the tiering already used for customer content — the '398 spec says so ("the control core may be considered to be an 'origin server' for the content for which it is authoritative"). MPEP § 2143 (B) simple substitution.

---

Summary table of grounds

---

5. Consolidated motivation rationale

Each of the seven MPEP § 2143 exemplary rationales has at least one concrete anchor in the '398 record:

• (A) Known elements combined by known methods, predictable results — DNS-based selection + HTTP object fetch + centralized configuration is the entire architecture of Farber/Ben-Shaul/'871.
• (B) Simple substitution — control core for origin server as the tier-0 source of URL-addressable objects.
• (C) Known technique improves similar device in the same way — Farber's DNS-based source selection is applied to internal control fetches exactly as it is applied to client steering; the spec says the two are "the same."
• (D) Known technique applied to a known device ready for improvement — '871's AMA/NOC broadcast-and-poll control plane, and Farber's contact-repeater table refresh, are the two structures the '398's GCO/CCS pull replicates.
• (E) Obvious to try — the design space (how do we get configuration to 1,000 caches?) had a small, well-explored set of answers; treating config as HTTP objects routable by DNS was one of them, with an evident expectation of success.
• (F) Design incentives / market forces — CDN scale, latency competition, and operating-cost reduction. Sandpiper's own complaint pleads the competitive pressure Level 3 faced ("Level 3 decided to exit the CDN market in 2023") — the incentive existed throughout.
• (G) Predictable variation — no performance threshold, algorithmic detail, or structural modification is claimed.

Additional doctrinal point — common ownership is not a shield. Because Farber '598 (issued 2001-02-06) and ATC '871 (issued 2010-10-26) both predate the § 102(b) critical date of 2010-12-14, they are § 102(b) art, not § 102(e)/(g) art, and pre-AIA § 103(c)(1) does not disqualify them. Likewise, under the AIA, the § 102(b)(2)(C) common-ownership exception reaches only § 102(a)(2) disclosures — inapplicable here. Patent Owner cannot escape its own predecessor's patents on a common-assignee theory.

---

6. Dependent claims — per-claim § 103 grounds

(Claim numbering and grouping are taken from the Prior-Art section's mapping; each is subject to the §0.4 provenance caveat. Dependent claims carry a presumption of obviousness once the independent claim falls unless they add a non-obvious feature — In re Fritch, 972 F.2d 1260, 1266 (Fed. Cir. 1992).)

---

7. Anticipated Patent Owner rebuttals and how to pre-empt them

---

8. Secondary considerations (Graham factor 4)

The record contains substantial commercial-success and copying evidence — Sandpiper's amended complaint pleads Microsoft's CDN revenue, Azure growth, and willfulness, and pleads Level 3's CDN leadership (https://storage.courtlistener.com/recap/gov.uscourts.txed.[238813](/patent/238813)/gov.uscourts.txed.238813.41.0.pdf). But the nexus is weak:

1. No claim-level nexus. The pleaded success (Azure, Office 365, Xbox, Windows Update traffic) is attributable to Microsoft's cloud business and to CDNs generally, not to the narrow control-plane architecture of claim 1. Nexus must be to the claimed invention, not to the product. In re GPAC Inc., 57 F.3d 1573, 1580 (Fed. Cir. 1995).
2. Portfolio-level pleading. The complaint's "value" evidence is portfolio- and history-based (Sandpiper Networks, the Starr Report, the 2006 sale, the "acquired the business primarily for its intellectual property" quote). That is not evidence of the non-obviousness of these particular claims.
3. Copying is equivocal. Sandpiper's copying theory is built on Microsoft's 2006 diligence on the Sandpiper assets and on later hires from CenturyLink/Savvis — i.e., knowledge of a portfolio, not duplication of the claimed control-plane design. And Microsoft's own accused CDN post-dates the asserted priority by years.
4. Long-felt need cuts the other way. The problem (scaling control-plane distribution) was identified and solved in the art by 1998–2002 — Farber, Ben-Shaul, Lewin. A need that was in fact met by the prior art cannot support non-obviousness.
5. No unexpected results, no teaching away, no skepticism identified in anything retrieved.

---

9. Practical constraints on running these grounds

1. § 325(d) is the real battleground. Several of the best references are already on the '398's or its family's face: Ben-Shaul US 2002/0010798 A1 and Lewin US 2002/0091801 A1 appear in the '398/family citation lists (Prior-Art section §2). Asserting them risks discretionary denial under Advanced Bionics / § 325(d) unless you can show the Office never applied them (the Prior-Art section could not establish which references were applied in the 2020-02-13 non-final or 2020-06-29 final rejections — verify this in the file wrapper first). The safer art set is the off-face art already being run against siblings: Glines, Gupta '252, Chung '389, Chiou '507, Wein '100, Liu US 2010/0257258, Devanneaux US 2007/0156845, Dilley/Pai/Wang.
2. The family is the roadmap. Google has instituted on Glines+Gupta (IPR2025-00826, '112) and on Dilley+Pai+Wang (IPR2025-00860, '573); Microsoft is running Lewin+Wein and Liu+Devanneaux against the same-specification '053 (IPR2026-00190). A '398 petition should be built by porting those Grounds onto the '398's claim language rather than starting from scratch.
3. § 101 may be the cheaper attack. The Litigation section records that siblings US 8,595,778 / US 8,719,886 were held § 101-invalid in district court. Claim 1 here is a functional apparatus claim over an architectural arrangement with no technical improvement recited — the Alice/Mayo step-two hook (no inventive concept beyond generic computer/DNS/HTTP elements) is at least as strong as the § 103 case, and § 101 is not available at the PTAB but is fully available in district court.
4. § 112(b) — "constructed and adapted to obtain CDN resources from at least some other CDN components" is a functional recitation with no structure or algorithm; it may be indefinite or purely functional under § 112(f) with no corresponding algorithm disclosed for the generic case.
5. No estoppel exists yet. Per the PTAB section, there is no AIA proceeding on the '398 — so § 315(e)(2) estoppel reaches nobody, and no § 315(b) clock runs on this patent until a complaint pleading it is served. That freedom cuts both ways: full art field, but also no safe harbor.

---

10. Bottom line

Claim 1 is very likely obvious under § 103. The combination of Farber US 6,185,598 + ATC US 7,822,871 + Ben-Shaul US 2002/0010798 A1 supplies every limitation, including the only contestable one (L4, reuse of the rendezvous mechanism to select an internal fetch source), because Farber already teaches a CDN-side element using DNS to select another CDN element and fetching its control tables from it, and Ben-Shaul already teaches edge components downloading policy-directive objects from a control server. The motivation is not reconstructed — it is recited in the patent's own specification, which states that control-core selection "may be made with the same rendezvous and selection mechanism(s) used to direct client requests."

The dependent claims add little. Each maps to art in the same field: tiered origin roles (Dilley/Pai/Wang — already instituted against a sibling), consensus/voting (the Paxos patents the spec itself names, plus Farber's succession election and two-phase commit), domain-name access to the control core ('871 + Farber), and cache-as-subscriber (Farber's Subscriber Table + Lewin).

The strongest defensive story for the Patent Owner is not novelty or non-obviousness — it is claim scope and nexus. If claim 1 is construed as broadly as the specification suggests ("CDN resources" = any URL-addressable object; rendezvous = any DNS/policy selector; "constructed and adapted to" = merely capable), then the claim is close to a generic description of a CDN, and the secondary-considerations evidence is portfolio-level rather than claim-level. Expect the Patent Owner to argue the reverse: narrow construction of "control core data" and "CDN resources," plus a nexus argument built on the commercial success of Level 3's and Microsoft's CDNs.

---

11. Confidence, uncertainty, and flagged contradictions

Confidence:

• High: that Farber '598, ATC '871, Ben-Shaul '798, Glines, Gupta '252, Chung '389, Chiou '507, Lewin '578, Wein '100, Liu US 2010/0257258 and Devanneaux US 2007/0156845 are patents/printed publications predating 2010-12-14 and thus § 102(b) art; that the PTAB instituted on Glines+Gupta ('112) and Dilley+Pai+Wang ('573); and that Microsoft is challenging the same-specification '053 over Lewin+Wein and Liu+Devanneaux.
• Moderate: the exact limitation-by-limitation mapping of Farber/Ben-Shaul to granted claim 1 (claim wording is reconstructed — §0.4); the pre-AIA vs. AIA determination (§0.1); that Farber's reflector counts as a "CDN component" of the same CDN as the repeaters in the claim's sense.
• Not established: the Dilley, Pai, and Wang patent/application numbers (cited here only as Ex. 1003/1004/1005 of IPR2025-00860); which references were applied (as opposed to listed) in the 2020-02-13 and 2020-06-29 rejections; and the exact granted text of the dependent claims.

Contradictions flagged against earlier sections:

1. Prior-Art §4, claims 16–17: the earlier section assigns "§ 102 exposure from self-disclosure" to "Applicant's own family (US 2014/0372588)." That appears to be wrong. Ser. No. 14/302,865 (US 2014/0372588 A1) is the '398's direct parent in the same family, with the same 2011-12-14 priority and the same inventive entity. Under pre-AIA § 102(e) a reference must be filed before the applicant's invention; under the AIA, § 102(a)(2) art must have an effective filing date before the claimed invention's. Neither is satisfied here. Family/same-priority disclosures are not § 102 prior art against the '398 unless the priority chain is broken. Flagging per your instruction; do not assert self-disclosure art without first proving a broken priority claim.
2. Priority date: the 2011-12-14 vs. 2011-12-13 discrepancy (Google Patents/spec vs. the Unified Patents portal snippet) remains unresolved. It is immaterial to every ground above — all principal references are years earlier — so I do not rely on the distinction.
3. Bench note (not a contradiction, but worth recording): the '398's single most damaging obviousness reference, US 6,185,598, is named to Farber and Swart — the founders of Sandpiper Networks, the "homage" company Sandpiper CDN, LLC names in its own complaint. The Patent Owner's narrative history (Sandpiper Networks' 1990s CDN innovation) therefore supplies, in the same breath, the prior art that most cleanly reads on claim 1's L3/L4.

Grounding URLs: grant — https://patents.google.com/patent/US10841398B2/en · Farber '598 — https://uspto.report/patent/grant/6185598 · ATC '871 — https://insight.rpxcorp.com/patent/US7822871B2 · Ben-Shaul '798 — https://patentimages.storage.googleapis.com/75/89/fa/aab384b72fddd3/US20020010798A1.pdf · Ben-Shaul grant US 6,976,090 — https://webapp1.dlib.indiana.edu/virtual_disk_library/index.cgi/5628977/FID979/OG/html/1301-2/us06976090-20051213.html · IPR2025-00826 petition analysis — https://ai-lab.exparte.com/case/ptab/IPR2025-00826/doc/1004 · IPR2026-00190 petition analysis — https://ai-lab-cl-prod.azurewebsites.net/case/ptab/IPR2026-00190/doc/summary/1 · IPR2025-00860 institution decision (Dilley/Pai/Wang) — https://www.docketalarm.com/cases/PTAB/IPR2026-00095/Microsoft_Corporation_v._Sandpiper_CDN_LLC/ · Sandpiper complaint / history — https://litigation.rpxcorp.com/news/80905-it-s-not-every-patent-complaint-that-name-checks-the-starr-report-and-big-bad-voodoo-daddy · https://storage.courtlistener.com/recap/gov.uscourts.txed.238813/gov.uscourts.txed.238813.41.0.pdf.