US 10708145

Summary of US Patent 10708145 (US10708145B2)

Bibliographic Data (verified via Google Patents / USPTO records)

Abstract (as stated in the patent record): "This invention relates to content delivery and content delivery networks. More specifically, to content delivery networks and systems, frameworks, devices and methods supporting content delivery and content delivery networks."

Technical Subject Matter (from the full specification)

The patent describes a service-oriented architecture for a content delivery network (CDN). Key concepts in the specification include:

• A taxonomy of CDN service types (delivery, configuration, control, event, and state information flows) and "minimal CD services" (control sinks + event sources).
• Interconnected service networks: configuration services, control services, collector services, reducer services, and primary delivery services (caching, rendezvous).
• Feedback loops: services emit event data that flows through reducers to collectors, which produce state information fed back to control services and to the originating services ("feedback from health service" per the title) for local and global control.
• Autonomic machine management ("Autognome"), layered service states, control manifests and control trees, and an origin-version-based invalidation mechanism for caches.

Independent Claims — Plain-Language Overview

I must flag a significant limitation: The claims portion of the patent was not included in the material available to me (the full-text excerpt I was given ends mid-specification, before the claims section), and my searches did not return the verified claim text or claim count for US10708145B2. Rather than risk fabricating or mischaracterizing the independent claims, I will not attempt a claim-by-claim plain-language breakdown. I cannot state with confidence how many independent claims the patent has or what their precise scope is.

CAFC 2026 Dockets

I searched for any CAFC 2026 docket involving patent 10708145 specifically and found no CAFC 2026 docket entries tied to this patent number. The related litigation that surfaced in searches involved Level 3 Communications in other matters (e.g., Level 3 v. Limelight, Fed. Cir. 2009-1480; the PersonalWeb/Twitch appeals, Fed. Cir. 20-1566/20-1568/20-1569 and 21-1858; and CAFC case 23-2181, dismissed by stipulation Feb. 8, 2024), but none of those identified US10708145 as a patent-in-suit. The Google Patents record for this patent does note "Family has litigation" and links to Darts-ip, but I could not verify any specific 2026 CAFC proceeding against this patent number from the searches performed.

Uncertainties

1. Claims: Not verifiable from available materials — no claim text was retrieved.
2. CAFC 2026 activity: No 2026 docket entries for this patent were found; absence of results is not conclusive proof that none exist, and PACER/docket databases may have records not surfaced by these searches.
3. Current assignee: The Google Patents record lists Sandpiper CDN, LLC as current assignee (recorded 2024-07-09) but notes assignee lists "may be inaccurate" — USPTO assignment records would be the authoritative confirmation.

Based on my searches, here is what I can confirm regarding litigation involving US Patent 10,708,145 ("Devices and methods supporting content delivery with adaptation services with feedback from health service"):

Ownership Context

The patent is currently assigned to Sandpiper CDN, LLC (assigned from Level 3 Communications, LLC in July 2024, per the Google Patents assignment record). Sandpiper CDN, LLC is the entity that has been asserting the former Level 3 CDN patent portfolio.

Direct Confirmation of 10,708,145 in a Specific Case

I could not confirm, from the search results returned, any specific district-court case or IPR in which US 10,708,145 is expressly identified as an asserted/challenged patent. The Google Patents record for the patent includes a "Family has litigation" indicator (with a Darts-ip link), so the patent family is in litigation, but I did not obtain docket exhibits affirmatively listing 10,708,145 in any particular matter.

Notably, in Sandpiper CDN, LLC v. [Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.) (E.D. Tex. 2:25-cv-00664, filed 6/26/2025), the amended complaint asserts six patents — 8,478,903; 8,924,466; 9,456,053; 9,228,883; 9,762,692; and 10,701,173 — not 10,708,145. (10,701,173 is a different patent; do not conflate the two.) Similarly, the complaint in Sandpiper CDN, LLC v. Microsoft Corp. (E.D. Tex. 2:26-cv-00681) shows exhibits A–H covering 9,021,112; 7,822,871; 7,860,964; 9,660,876; 9,647,899; 10,116,738; 8,156,066; and 8,645,517 — again without 10,708,145 visible in the excerpted exhibit list.

Related Sandpiper CDN Litigation Campaign (same family/portfolio, for reference)

These cases involve Sandpiper CDN, LLC (the current owner of 10,708,145) but I cannot confirm 10,708,145 is asserted in them:

Related IPRs (Petitioner challenges to Sandpiper CDN patents)

• Google LLC v. Sandpiper CDN, LLC: IPR2025-00806, -00826, -00860, -00969, -01010 (filed 2025; some petitions referred per papers 13/14).
• Microsoft Corporation v. Sandpiper CDN, LLC: IPR2026-00095 (filed 11/3/2025, pending; POPR filed 2/13/2026); IPR2026-00174 (filed 12/17/2025, pending; confirmed by RPX as an IPR of US 9,762,692 — not 10,708,145); IPR2026-00180 (filed 12/24/2025, pending).

Bottom Line

• No case specifically identifying US 10,708,145 as an asserted patent could be confirmed from the searches performed.
• The patent's owner (Sandpiper CDN, LLC) is actively litigating the broader Level 3 CDN portfolio against Google, Comcast, and Microsoft, and the Google Patents record flags the family as having litigation.
• To definitively identify 10,708,145's involvement, you should check PACER/Unified Patents/RPX for the specific complaints (full exhibit lists) and the PTAB filings — the docket excerpts I retrieved did not list 10,708,145, but truncated exhibit lists could miss it.

I am stating this with the caveat that my search reached its step limit, so I could not exhaustively verify every docket's full patent list.

Proceedings overview

Zero AIA trial proceedings are on file for US10708145. The USPTO Open Data Portal returns no IPR/PGR/CBM proceedings for this patent, and targeted web searches (including searches pairing "10708145" with "IPR," "final written decision," and "PTAB") surfaced no petition, institution decision, or Final Written Decision directed to this patent. The breakdown is: 0 active, 0 with claims invalidated, 0 with claims sustained, 0 settled, 0 institution-denied. The defensive posture is therefore "untested and unhardened" — every claim remains in force, but no estoppel binds any challenger, and the prior-art field is wide open for a first mover.

Note on near-miss proceedings (flagging, not fabricating): Web search did surface an active PTAB campaign by Microsoft Corporation against Sandpiper CDN, LLC — the current assignee of this patent — but none of those proceedings challenges US10708145. Specifically:

• IPR2026-00095 (Microsoft v. Sandpiper CDN, filed 2025-11-03, joinder with Google's IPR2025-00969) challenges US 8,478,903 — a different patent in the same Level 3/Sandpiper CDN portfolio. Panel: APJs Mitchell G. Weatherly, Sheila F. McShane, Michael T. Cygan (per DocketAlarm). The Google IPR was instituted on the merits 2025-11-28; Microsoft's copycat joinder petition was pending with institution decisions in the May–July 2026 window.
• Microsoft's related filings — IPR2026-00174 (US 9,762,692), IPR2026-00180 (US 10,701,173), IPR2026-00190 (US 9,456,053) — likewise target sibling patents, not this one.
• These arise out of Sandpiper CDN, LLC v. Microsoft Corp., No. 2:25-cv-00664 (E.D. Tex.), where Sandpiper asserts six patents (8,478,903; 8,924,466; 9,456,053; 9,762,692; 10,701,173; and another) — US10708145 is not on the asserted list in that case, nor was it among the six patents Sandpiper asserted against Google in 2:24-cv-03951 (C.D. Cal.) per RPX reporting (8,478,903; 8,595,778; 8,645,517; 8,719,886; 9,021,112; 10,924,573).

Strategic summary

Claim status. Every claim of US10708145 is UNTESTED in AIA trial proceedings. No PTAB proceeding has ever challenged, much less canceled, any claim. The patent issued 2020-07-07 from an application filed 2012-12-14, it is assigned to Sandpiper CDN, LLC (via Level 3 Communications), and its adjusted expiration is 2035-09-14 — meaning the full claim set is presently enforceable and will be for roughly another nine years. There is no FWD to quote, no canceled claim to rely on, and no PTAB-derived narrowing. If a demand letter cites US10708145, there is no "dead claim" argument available from PTAB records.

Estoppel landscape. Because no IPR/PGR/CBM has been instituted on this patent, no § 315(e)(2) estoppel attaches to anyone with respect to US10708145. No petitioner has raised or "reasonably could have raised" any ground against it, so every available prior-art ground — § 102, § 103, and § 112 challenges — remains available to any defendant. The flip side is that the patent owner faces no estoppel either, and can assert the full claim set without any PTAB-narrowed construction. One strategic caution: if a defendant is also party to (or in privity with) the Microsoft or Google IPRs on the sibling patents, that estoppel only runs to those specific patents — it does not extend to US10708145, which was never challenged.

Pattern signals. Several patterns are worth reading:

• Same plaintiff, same portfolio, different patents. Sandpiper CDN (a Delaware NPE formed 2024-03-21, which took ~80+ Level 3 patents via an April 2024 assignment — itself a sale following Level 3's 2023 CDN exit) has sued Google (C.D. Cal.), Comcast (E.D. Tex.), and Microsoft (E.D. Tex.) on the broader CDN family. US10708145 is squarely within that family (same inventors — Lipstone, Crowder, Swart, Newton, Varney — same filing date, same CDN subject matter, and it shares a Google Patents family marker with the litigated patents). The fact that Sandpiper has not asserted it yet may simply mean it is being held in reserve for a second wave — or that the plaintiff regards it as weaker than the siblings already in suit.
• Defensive-aggregator and defendant activity is concentrated on siblings. RPX has covered the campaign; Microsoft has shown it will file copycat/joinder IPRs on every patent asserted against it (four IPRs in Nov–Dec 2025). If US10708145 gets asserted, a Microsoft-style response — petition plus joinder where feasible — is the established playbook. Unified Patents does not appear in the chain for this patent specifically.
• Patent owner fights. Sandpiper has filed discretionary-denial requests (e.g., Fintiv arguments) and Director Review requests in the sibling IPRs — it is litigating the PTAB stage aggressively, so any future IPR on this patent should expect a hard-fought institution phase.

Recommended next steps

• No claims are canceled — do not premise a defense on PTAB invalidity. If you are a defendant facing US10708145, the PTAB record gives you nothing to cite. The honest position: this patent has never been tested, so your invalidity case must be built from scratch — prior-art searches (the 1990s Sandpiper-era art used against the '903 patent, e.g., Kenner WO 96/41285, is a good starting analog), prosecution-history review of the 13/715,466 file, and § 101 subject-matter eligibility analysis for the "adaptation services with feedback from health service" claim architecture.
• Monitor the sibling IPRs as a leading indicator. Institution decisions in IPR2026-00095 / -00174 / -00180 / -00190 were due May–July 2026. The art and arguments developed there — especially any art that reads on the shared CDN architecture — are publicly available and highly probative for a future challenge to US10708145, even though estoppel will not carry over. Watch Google's IPR2025-00969 (instituted 2025-11-28) for the first substantive PTAB analysis of this family's core claims.
• No active proceeding exists on this patent, so there are no PTAB milestones to calendar. The absence of any IPR in the ~6 years since issuance is itself a signal: this patent has apparently never been the subject of serious, asserted commercial activity (it was not in the Google, Comcast, or Microsoft complaints), and well-asserted patents in this family (the '903 patent) drew IPRs within months of assertion. If you receive a demand letter citing US10708145, treat it as the first real test of the patent — and note that a petition would face no § 315(e)(2) bar, no prior PTAB constructions, and a patent owner with demonstrated appetite for discretionary-denial and Director Review fights.
• Verify the ODP gap yourself before filing anything. The structured data reflects the most recent ODP ingest; check PTAB E2E (https://ptab.uspto.gov) and the USPTO Patent Center for US10708145 immediately before any deadline, since a petition filed after the ingest date would not appear in the API feed.

Bottom line: US10708145 is a virgin target — all claims live, zero PTAB history, no estoppel against anyone. A defendant facing it cannot point to any IPR victory, but equally faces no IPR-derived obstacles: the first challenger gets a clean slate on every statutory ground.

Based on my research (USPTO/Google Patents legal events, RPX Insight coverage of the Sandpiper CDN campaign, PTAB filings in IPR2026-00190, and district-court dockets), here is the reconstructed assignment record and NPE analysis.

Inventors

Pattern note: All five inventors assigned their interests to Level 3 Communications, LLC in March 2013. The notable pattern here is not post-filing departure but lineage: the 2012 application traces back through the Sandpiper Networks → Digital Island → Savvis → Level 3 acquisition chain, and the patent ultimately landed (2024) at an entity named after Sandpiper Networks — a full-circle monetization structure. Andrew Swart's name appears both on the 1990s Sandpiper patents and on this 2012 filing.

Original assignee

Level 3 Communications, LLC (named on the issued patent; the LLC was the operating subsidiary of Level 3 Communications, Inc.).

• Product line: Level 3 operated one of the largest commercial CDNs in the US; this patent (CDN architecture with adaptation services fed by health-service feedback) is squarely within the CDN platform Level 3 operated. It shipped products embodying the claimed subject matter.
• Status: Acquired by CenturyLink in 2017 (now Lumen Technologies); Level 3 exited the CDN business in 2023 and began selling CDN assets, including this patent, in 2024 (per Sandpiper's own complaints). Not bankrupt; a strategic-exit sale.

Assignment timeline

The USPTO Assignment Center record for this patent (as corroborated by a PTAB exhibit and RPX) shows two recorded links. I could not retrieve the correspondent names or the 2013 reel/frame from available sources — those fields are flagged below rather than guessed.

• 2012-12-14 (executed with filing) / recorded 2013-03-26 — Reel/frame not retrievable from sources used (Google Patents legal event confirms recording date only)

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Andrew Swart, Laurence R. Lipstone, William Crowder, Christopher Newton, Lewis Robert Varney
  • Assignee: Level 3 Communications, LLC
  • Correspondent: not determinable from sources used
  • Context: Standard inventor-to-employer assignment on the original filing.

• 2024-04-24 (executed) / recorded 2024-04-26 — Reel 068256 / Frames 0091–0115

  • Conveyance: Assignment of Assignors' Interest (portfolio assignment of 80+ US patents)
  • Assignor: Level 3 Communications, LLC (then a Lumen Technologies subsidiary)
  • Assignee: Sandpiper CDN, LLC (Delaware LLC, formed 2024-03-21)
  • Correspondent: not determinable from sources used; flag for verification — if the same recording attorney appears on other Sandpiper-family reels, that is the repeat-correspondent tell.
  • Context: Transfer-to-asserter / exit monetization — Level 3's CDN patent portfolio sold to a newly formed Delaware LLC that began suing within two weeks of recording.
  • Sources: RPX Insight (assignment "dated April 24, 2024, recorded with the USPTO on April 26"); PTAB IPR2026-00190, Petitioner Exhibit EX1036 ("Patent Assignment Cover Sheet from Level 3 Communications, LLC to Sandpiper CDN, LLC, Reel No. 068256, Frames 0091-115"). Google Patents' legal-events tab lists the Sandpiper assignment event under 2024-07-09 (likely an ingestion/publication date for the same or a companion record).

No other recorded assignments found for US10708145. There is no record of this patent passing through the 2017 Optic153/Equitable IP transfer (that batch predated issuance of this 2020-granted patent).

Timeline diagram

timeline
    title Ownership of US 10708145
    2012 : Filed by Level 3 Communications
    2013 : Inventors assign to Level 3 LLC
    2017 : CenturyLink acquires Level 3
    2020 : Patent issued
    2024 : Assigned to Sandpiper CDN LLC
         : Sandpiper sues Google

NPE / troll-pattern signals

1. Shell-entity transfer — present. Level 3 (operating CDN) transferred this patent, inside an 80+ patent portfolio, to Sandpiper CDN, LLC, a Delaware LLC formed 2024-03-21, with no CDN products or operations — an assertion vehicle. Tells: single-purpose Delaware LLC, formed weeks before the transfer, no commercial operations, immediate litigation. Evidence: RPX Insight, May 12, 2024; Reel 068256/Frames 0091-115 (recorded 2024-04-26).

2. Known asserter in the chain — present. Sandpiper CDN, LLC is a tracked NPE plaintiff: RPX Insight describes it as a "recently formed Delaware plaintiff" running a CDN-patent litigation campaign. Dockets: Sandpiper CDN, LLC v. Google LLC, 2:24-cv-03951 (C.D. Cal., filed 2024-05-10); Sandpiper CDN, LLC v. Comcast Cable Communications LLC, 2:24-cv-00886 (E.D. Tex., filed 2024-11-01); Sandpiper CDN, LLC v. [Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.), 2:25-cv-00664 (E.D. Tex., filed 2025-06-26). Note: US10708145 is held in the asserted portfolio but does not appear on the published asserted-patent lists in those three cases (Google: '903/'778/'517/'886/'112/'573; Microsoft amended complaint Exhibits A–F; Comcast: '347/'876 plus others) — it is portfolio inventory, not yet a named patent-in-suit in a case I could verify. RPX also notes Level 3's prior transfer of ~110 patents to NPE Optic153 LLC / Equitable IP Corp (2017), a same-family pattern.

3. Repeat correspondent across the chain — unclear. I could not retrieve correspondent names for either recording from the sources available; do not fabricate. If the same attorney appears on the 2013 and 2024 cover sheets and on other Sandpiper reels, that would confirm the tell — verify at the Assignment Center link below.

4. Cascading transfers — not present. Only two links (2013, 2024) with an 11-year gap; no chained LLCs.

5. Pre-litigation transfer — present. Assignment executed 2024-04-24 and recorded 2024-04-26 (Reel 068256); Sandpiper sued Google 2024-05-10 — ~14 days after recording, comfortably inside the 6-month window. The chain was arranged to enable assertion with a clean standing record.

6. Bankruptcy fire-sale — not present. No bankruptcy; Level 3 was acquired by CenturyLink/Lumen in 2017 and sold the CDN patents as a deliberate exit strategy.

7. Privateering — unclear. The transferor (Lumen via Level 3) sold, rather than licensed/retained an interest, and had exited the CDN market; classic privateering (operating company retains economics while an NPE sues its competitors) is not documented in public SEC filings I could verify. The behavior — an operating company's CDN patents asserted against CDN operators (Google, Comcast, Microsoft) — is consistent with privateering, but the retained-interest evidence is missing.

8. Defensive aggregator (anti-NPE) — not present. The chain terminates at Sandpiper CDN, LLC, an asserting entity — not RPX, AST, LOT, Unified Patents, or OIN.

Verdict

NPE — high confidence.

Two-plus strong signals are documented, not inferred: (1) a shell-entity transfer — Level 3 moved this patent to Sandpiper CDN, LLC, a Delaware LLC formed on 2024-03-21 with no products, per RPX and the PTAB-submitted cover sheet (Reel 068256, Frames 0091-115, recorded 2024-04-26); and (2) a pre-litigation transfer to a known asserter — Sandpiper filed its first CDN-patent suit against Google on 2024-05-10, two weeks after recording, and is tracked by RPX as an NPE plaintiff. The remaining uncertainty is limited to the correspondent identity and whether this specific patent is on a current complaint, neither of which changes the portfolio-level conclusion.

Verification: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search Patent No. 10708145; key record: Reel 068256/Frames 0091-115).

I'll research the cited prior art for US Patent 10,708,145 B2. Let me start by searching for the patent's citation records.

Let me dig deeper into the specific citations listed for this patent on Google Patents.

Let me try more targeted searches to find the exact citation list.

I have to stop searching at this point, so I'll give you a transparent assessment of what could and could not be verified, and provide the most useful analysis possible without fabricating data.

Important caveat up front

I could not retrieve the "References Cited" (prior-art citations) section of US 10,708,145 B2 from any source available to me. Here's what I verified and what I could not:

• Verified: The bibliographic record (patent number, title, application 13/715,466, priority date Dec. 13, 2012, issue date July 7, 2020, Level 3 Communications → Sandpiper CDN, LLC assignee history). The full specification you supplied is authoritative for the disclosure, and I have reviewed it.
• Not retrievable: (a) the claims of the patent — the specification text you provided ends mid-sentence before the claims section, and my searches did not surface the verified claim text or claim count; and (b) the examiner's list of cited references (the "References Cited" section that appears on the face of an issued patent, and in the USPTO's citation/IDS records). My searches surfaced the patent's bibliographic record, its family members (EP2932401B1, CA2894873C, WO2014093717A1, HK1215817B), and pages of other patents that cite or reference this patent — but not the list of references this patent cites.

Because the claims and the citation list are both unavailable, I cannot truthfully produce a claim-by-claim § 102 anticipation map, and I will not invent one. Any list of "cited references" I fabricated would be a serious analytical error. What I can do is (1) tell you exactly where to get the authoritative citation list, and (2) describe the prior-art landscape that is genuinely relevant to this patent's subject matter, clearly labeled as contextual rather than as verified examiner citations.

Where to obtain the authoritative citation list

The front page of the issued patent (the USPTO PDF) contains the examiner's "References Cited" section, including U.S. patent documents, foreign patent documents, and non-patent literature. Authoritative sources:

1. USPTO Patent Center – patentcenter.uspto.gov (application number 13/715,466); the "Citations" and "IDS" tabs and the image file wrapper will show the cited art and the § 102/103 rejections made during prosecution.
2. Google Patents – patents.google.com/patent/US10708145B2/en, "Citations" tab (my search queries did not return the contents of that tab, but it is on the live page).
3. USPTO PatFT/Patent Public Search – the front-page "References Cited" field.
4. Global Dossier (via the Google Patents record for this patent) for the foreign family counterparts, which may share the same cited art.

Verified relevant context: the disclosure's subject matter (for prior-art framing)

Based on the specification you provided, the patent claims priority to a December 13, 2012 filing and covers, at a high level:

• A service-oriented CDN architecture with categorized service types (delivery, configuration, control, events, state) and "minimal CD services" (control sinks + event sources).
• Feedback loops from a "health service"/collector/reducer network: services emit event streams → reducers → collectors → state information fed back to control services and back to the originating services for local/global control (matching the title, "feedback from health service").
• Autonomic machine provisioning ("Autognome"), layered service states (Level 0–4), control manifests/control trees, and origin-version-based invalidation for caches.

For § 102 (novelty) purposes, the most pertinent prior art would be anything published before December 13, 2012 disclosing: (i) CDN service frameworks with control/event/state feedback loops, (ii) autonomic/self-configuring network services, or (iii) origin-version-based cache invalidation.

Prior-art landscape relevant to this subject matter (contextual — NOT verified as examiner citations)

The following are well-known references in the CDN art that a competent search for this disclosure would surface. I have not confirmed that any of these appear on the face of US10708145B2, and I cannot map any of them to specific claims without the claim text. Treat these as candidate art, not as the patent's actual citation list:

Again: this table is contextual guidance, not a verified citation list. I have not seen the examiner's citations for this patent, and I cannot state which references appear in the "References Cited" section or which claims any reference would anticipate.

On the § 102 anticipation question

A proper § 102 anticipation analysis requires:

1. The claim text of US10708145B2 (to map each limitation), and
2. The cited reference (to show each element is disclosed, arranged as in the claim, by a single prior-art reference).

Since neither the claims nor the citation list were retrievable in my searches, I cannot identify with confidence which claim(s) any specific reference "potentially anticipates." I will not guess, because mislabeling a reference as anticipating a specific claim number would be worse than saying I don't know.

Bottom line

• Verified: The patent's bibliographic identity (US 10,708,145 B2, app. 13/715,466, priority Dec. 13, 2012, issued July 7, 2020, Level 3/Sandpiper CDN) and its disclosure (service-oriented CDN with health-feedback loops, Autognome, control manifests, origin-version invalidation).
• Not verified and needed next: the claims and the "References Cited" list. Pull the USPTO PDF front page or Patent Center records; those will give you the exact cited art.
• I have not fabricated a citation list or claim-to-reference mappings. If you can provide the patent's front page (References Cited) or the claims, I can immediately produce the full citation-by-citation analysis you're asking for, with publication dates, descriptions, and § 102 mapping.

I'll research the prior art landscape for this patent. Let me first locate the citations listed on the patent face and related references.

The initial searches didn't surface the examiner-cited references for this specific patent. Let me look deeper at the patent's citation list and its related family.

Let me access the full citation lists from the application record and the granted patent.

I have reached the search step limit, so I'll now provide the obviousness analysis based on the available verified materials and well-established prior art. I want to be transparent about an important evidentiary constraint first.

Obviousness Analysis Under 35 U.S.C. § 103 — US 10,708,145 B2

1. Threshold Limitations (read before relying on this analysis)

Two structural limitations affect this analysis, and I flag them explicitly rather than papering over them:

1. The claims are not available in the material provided. The full-text excerpt supplied ends mid-specification, before the "Claims" section, and my searches did not return verified claim text for US10708145B2. A rigorous § 103 analysis must be anchored to claim limitations; without claim text, I cannot map prior art element-by-element to the granted claims. What follows is therefore an obviousness analysis at the level of the disclosed invention concepts (which is the best available proxy), and it should be treated as a prima facie framework, not a definitive claim chart.

2. The "Prior Art section of this page" referenced in the task was not included in the message, and I could not retrieve the examiner's cited-references list for this specific patent before hitting the search limit. I verified several references indirectly through the patent's own family (sibling applications list overlapping prior art), and I identify below which references are verified (appeared in search results tied to the Level 3 2012–2013 family) versus which are well-known art from training data that a person of ordinary skill in the art (PHOSITA) would indisputably know. Where I could not verify, I say so.

Prosecution signal worth noting: Search results show U.S. Appl. No. 13/715,466 received a Final Office Action (Dec. 21, 2015) and an Advisory Action (Jan. 13, 2016) before issuance in 2020 — i.e., the granted claims were narrowed/reworked during prosecution, so the issued claims are almost certainly narrower than the specification's broad disclosure. That makes the "concept-level" obviousness mapping below a starting point; the actual § 103 fight would be over the specific claim limitations added to overcome those rejections.

2. The Person of Ordinary Skill in the Art (PHOSITA)

A PHOSITA for this patent would be a team with (i) a CDN architect's knowledge of request routing (DNS/rendezvous), caching, and origin fill; (ii) a distributed-systems engineer's knowledge of configuration management, monitoring, event/stream processing, and control loops; and (iii) working familiarity with IETF standards (HTTP/RFC 2616 caching semantics, DNS), autonomic-computing literature, and the published CDN literature (Akamai, Digital Island, Limelight). Skill level: MS/PhD or equivalent industry experience, ~5+ years in content delivery or large-scale distributed services as of December 2012.

3. Representative Prior Art Landscape (verified where possible)

4. Obviousness Combinations by Disclosed Concept

The patent's specification (supplied) centers on: (a) a service taxonomy (delivery/config/control/event/state information flows; "minimal CD service" = control sink + event source); (b) feedback loops from services → reducers → collectors → state → control (the "health service" feedback); (c) Autognome autonomic service provisioning with layered states; (d) control manifests/control trees for distributing configuration; and (e) origin-version invalidation (mov/bov/rov). Below are the combinations a challenger would most likely assert, with motivation analysis.

Combination A — "CDN service-oriented control architecture with feedback" (concepts: service taxonomy, minimal CD service, control/event/state flows)

• Primary: Newton et al. US 2013/0159472 (Level 3) — already discloses a CDN with a control core distributing global config to caches and customers.
• Secondary: Dilley et al. 2002 + Leighton US 6,108,703 — disclose CDN infrastructure services (DNS/rendezvous, edge caches, monitoring/measurement) and health-based request routing; Kephart & Chess 2003 — discloses categorizing managed elements by information flows and closing a monitor-analyze-execute loop; Lipstone US 2012/0215779 — discloses CDN analytics collection.
• Mapping: Every CDN service consuming control and emitting event data is a routine application of the MAPE-K loop to a known CDN. The "five information categories" table (Table 1) is a taxonomy — mere organization of known categories (config, control, events, state, delivery) each of which individually existed in the prior art (config = NETCONF/manifests; control = CDN control planes; events = syslog/NetFlow; state = monitoring DBs; delivery = HTTP).
• Motivation: A PHOSITA designing a next-generation CDN in 2012 would have had an obvious, market-driven incentive to formalize the ad hoc monitoring/control mechanisms of Akamai/Digital Island-era CDNs into a uniform service framework to reduce integration cost and enable closed-loop management. The references themselves point to the combination: Kephart & Chess expressly argues autonomic elements must expose sensor/effector interfaces (control in, events out) — exactly the "minimal CD service."

Combination B — "Health-service feedback via reducers and collectors" (concepts: reducer/collector networks, state feedback to control and to the originating service)

• Primary: Lipstone US 2012/0215779 (analytics management in CDN) + Farber US 8,296,396 (rendezvous using load/network-condition feedback).
• Secondary: Apache Scribe/Chukwa/Flume (hierarchical log reduction) and Twitter Storm (2011) (stream reduction), both well-known by 2012; Farber US 7,949,779/7,945,693 (rate control from collected subscriber data).
• Mapping: "Services emit events → reducers aggregate → collectors produce state → control services adapt delivery" is the standard telemetry → analytics → policy pipeline present in every commercial CDN and in the MAPE-K loop. The patent's Fig. 1K "localized feedback" (state fed back to the originating service endpoint) is the classic closed-loop controller.
• Motivation: The prior art expressly teaches why: Farber 8,296,396 routes around unhealthy/loaded servers; Lipstone 2012/0215779 aggregates CDN analytics; rate-control patents (7,949,779/7,945,693) throttle using measured subscriber information. Combining a reducer/collector data plane with a control plane is the predictable application of known aggregation technology to a known problem (CDN health monitoring), yielding the predictable result of network-wide adaptive behavior. Under KSR, this is the paradigm case of "the combination of familiar elements according to known methods… [yielding] a predictable result."

Combination C — "Autognome autonomic service provisioning" (concepts: layered machine states, service constellations, autonomic install/start/probe/stop)

• Primary: Kephart & Chess 2003 (MAPE-K; self-configuration) + NetServ (Columbia, 2011) — a service framework that dynamically instantiates services on network nodes with versioned bundles.
• Secondary: cfengine/Puppet-style desired-state configuration management and OS package managers (yum/apt), both ubiquitous by 2012.
• Mapping: Autognome's install→start→probe→stop cycle with per-service state machines is a specialization of (i) autonomic manager elements and (ii) package-manager state transitions (installed/running/stopped). "Level 0–4" layering (OS kernel → Autognome self-config → service provisioning → dynamic service config) maps to the well-known separation between immutable platform, bootstrap agent, and dynamically managed application config.
• Motivation: The dominant problem in CDN operations by 2010–2012 was fleet-scale configuration drift; the literature (Kephart & Chess; Puppet/cfengine documentation) expressly motivated autonomous, desired-state-driven configuration to eliminate per-machine human intervention. NetServ provided a working in-network proof that services could be dynamically installed on network nodes. Combining a CDN (Newton 2013/0159472) with autonomic provisioning (Kephart & Chess + NetServ) is an obvious "apply known technique to a known environment" (§ 103 KSR) — no inventive leap.

Combination D — "Control manifests/control trees + origin-version invalidation" (concepts: control resources, invalidation manifests, mov/bov/rov)

• Primary: Farber US 8,061,613 (resource invalidation in a CDN) + Newton US 2013/0159472 (CDN config distribution).
• Secondary: RFC 2616 HTTP cache-invalidation/revalidation semantics; RFC 4741 NETCONF (config tree distribution and subscription); Level 3 family disclosures on invalidation sequencing (e.g., 2014/0344452 "Invalidation Sequencing," same inventor group).
• Mapping: The specification's invalidation section (origin "minimum origin version," "latest origin version," stale-when-rov < mov) is a version-timestamp invalidation scheme. Farber 8,061,613 already teaches origin-based invalidation in a CDN; HTTP caching has taught freshness/expiration for two decades. The "control manifest as a hierarchical dictionary with reference slots (@, @@, %)" is a configuration tree — the same structure NETCONF and countless config-template systems (e.g., Apache/Tomcat configuration hierarchies) use.
• Motivation: A PHOSITA combining Farber's invalidation with Newton's control distribution would have been motivated by the well-recognized need to invalidate both cached content (client-facing) and control state (service-facing) — the specification itself concedes these are two known mechanisms ("cache-invalidation and other policies defined by the HTTP protocol… the basic control notification mechanism"). Transporting invalidation commands over the control/notification channel is an obvious design choice where a control channel already exists.

Combination E — "Rendezvous/adaptation using health feedback" (title concept: "adaptation services with feedback from health service")

• Primary: Farber US 8,296,396 (rendezvous based on load balancing and network conditions) + Major US 8,868,772 (adaptive-rate shifting) + LiveSky (Yin et al. 2009, hybrid CDN-P2P live streaming with adaptation).
• Secondary: HLS/Smooth Streaming/DASH adaptive-bitrate practice (2011–2012); Farber US 7,949,779/7,945,693.
• Mapping: Adapting delivery (bitrate, server selection, peer selection) in response to health/load telemetry was fully established: DNS-based routing on health metrics (Farber 8,296,396), rate adaptation (Major; HLS), and hybrid CDN-P2P adaptation (LiveSky).
• Motivation: The express motivation is in the references themselves — LiveSky describes adapting delivery paths based on observed network conditions; Farber routes based on server load. Combining "health service" telemetry with "adaptation" (rendezvous/rate selection) is the textbook feedback control problem, and no unexpected result would have been predicted.

5. Graham / KSR Framework Applied

• Scope and content of prior art: As above — CDN routing/invalidation (Farber, Leighton), control planes (Newton), analytics (Lipstone), autonomic frameworks (Kephart & Chess, NetServ), streaming adaptation (Major, LiveSky, HLS). All pre-date the Dec. 13, 2012 priority date.
• Differences: The specification's differences are chiefly (i) a formal taxonomy of service information flows, (ii) uniform treatment of all services as control sinks/event sources, and (iii) specific reducer/collector/control topology with local feedback. These are design-architecture choices, not new physical mechanisms.
• Level of ordinary skill: High — CDN/distributed-systems engineers in 2012 routinely built exactly these pipelines.
• KSR factors favoring obviousness: (a) Predictable variation — combining known telemetry, known control, and known adaptation; (b) Design incentive / market pressure — the 2010–2012 CDN market (Akamai, Limelight, EdgeCast, CDNetworks) competed on control-plane sophistication; (c) Known problem with known solution — fleet-scale configuration and health feedback were long-solved in autonomic-computing literature; (d) Teaching/suggestion/motivation in the references — MAPE-K and NetServ expressly teach service frameworks with control/event interfaces; the CDN references expressly teach health-based routing.
• Secondary considerations: Unknown from the available record. No objective indicia (long-felt need, commercial success, copying, unexpected results, licensing) were verifiable. If litigation (the "family has litigation" flag) later surfaces evidence of long-felt need or industry copying, that could rebut prima facie obviousness — but nothing in the provided record supports it.

6. Honest Assessment of Each Combination's Strength

Most dangerous single combination overall: Newton 2013/0159472 (Level 3's own CDN control-core disclosure) + Farber 8,296,396 (health-based rendezvous) + Lipstone 2012/0215779 (CDN analytics) + Kephart & Chess (autonomic loop) — together these cover, at concept level, a CDN in which services emit events, analytics are reduced/collected, state feeds back to control and to the service itself, and delivery is adapted. This is the combination a § 103 challenger would lead with.

7. Conclusion

On the available record, a strong prima facie § 103 case exists against the disclosed concepts of US10708145B2 using combinations of: (1) the verified Level 3/Digital Island family art (Newton, Farber, Lipstone, Lisiecki), (2) the Akamai architecture literature (Leighton/Dilley), (3) autonomic-computing prior art (Kephart & Chess; NetServ), and (4) standard CDN/streaming adaptation references (Major, LiveSky, HLS). The motivation to combine is supplied by the well-documented industry problems of fleet-scale configuration, health-based routing, and closed-loop adaptation, and by the references' own teachings.

However, this conclusion is provisional. Because the granted claim text was not available in the provided material (and the examiner's cited references for US10708145B2 were not retrieved before the search limit), I cannot certify that any particular combination reads on the granted claims. The prosecution history (Final OA Dec. 2015) indicates the claims were narrowed; the decisive question is whether the surviving limitations — e.g., the specific "minimal CD service" definition, the two-stage reducer→collector→state-feedback topology, the layered service-state hierarchy, or the mov/bov/rov invalidation formalism — are present in the single most relevant references (Newton 2013/0159472, Farber 8,061,613 and 8,296,396) or were added only to overcome these very rejections. Obtaining the granted claims and the examiner's § 103 rejections from the file wrapper (Public PAIR/Patent Center) is the necessary next step before this analysis can be finalized.