Obviousness

To analyze the obviousness of US patent 10,020,961 under 35 U.S.C. § 103, we need to identify combinations of prior art references that would render the claims obvious and explain the motivation for combining them. The provided patent text lists "Prior art keywords" as "tunnel, packet, domain, end point, switching." It also explicitly mentions "VXLAN (virtual extensible local area network)" and "NVGRE (network virtualization using generic routing encapsulation)" as "Most representative overlay-based network virtualization" and highlights their limitations as the problem the invention aims to solve.

Since a specific list of prior art references applied by the examiner during prosecution is not directly provided in the text for detailed analysis, I will generally discuss how common knowledge in the field and the identified "Prior art keywords" could be combined, focusing on the problem US10020961 seeks to address: the expandability limitations of existing overlay-based virtual networks like VXLAN and NVGRE, particularly in multi-domain scenarios.

General Principles of Obviousness (35 U.S.C. § 103):

A patent claim is obvious if "the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains." This involves considering:

1. Scope and content of the prior art: What was publicly known before the patent's priority date (December 27, 2013, for US10020961)?
2. Differences between the prior art and the claims at issue: What unique features does the patent introduce?
3. Level of ordinary skill in the pertinent art: The capabilities of a hypothetical person working in the field of network virtualization.
4. Secondary considerations (e.g., commercial success, long-felt need, failure of others): These can provide objective evidence of non-obviousness.

Motivation to Combine:

For a combination of prior art references to render a claim obvious, there must be a discernible reason, suggestion, or motivation for a person of ordinary skill in the art to combine those references in the way claimed by the invention. This motivation can come from:

• The prior art references themselves (e.g., one reference explicitly suggesting combining with another type of system).
• Knowledge of a person of ordinary skill in the art (e.g., known techniques for overcoming a particular problem).
• The nature of the problem to be solved (e.g., combining elements to achieve a desired function or improvement).

Analysis of Obviousness for US10020961:

The patent aims to provide expandability in overlay-based virtual networks, particularly addressing the limitations of VXLAN and NVGRE, which are restricted in size due to the N*(N-1) tunnel requirement for full-mesh topology in a single domain. The solution proposed is to introduce "domain tunnel end points" to connect multiple "network virtualization domains" in a full-mesh topology, thereby forming a larger virtual network. Each domain tunnel end point acts as a "tunnel switch" to relay packets between edge tunnel end points that are not directly connected via a single-hop tunnel (Description, "FIG. 1").

Let's consider the elements of the independent claims (Claim 1, 7, and 13) and how they might relate to existing prior art and the motivation to combine.

Elements of the Claims:

• Tunnel Manager: Collects tunnel end point information and connects tunnels.
• Tunnel Packet End Point: Receives tunnel packets, processes for L2 switching, transmits to domain VSI.
• Domain VSI (Virtual Switching Instance): Performs L2 switching on processed packets, transmits to tunnel packet generator. The patent also details a modified flooding rule within the domain VSI (Claim 2, 8).
• Tunnel Packet Generator: Adds tunnel header (including VN header and L3 tunnel header) to L2-switched packets to create and transmit new tunnel packets (unicast in Claim 13).
• Multi-domain structure: Generating at least two network virtualization domains, each with a domain tunnel end point and edge tunnel end points. Domain tunnel end points are connected with L3 tunnels in full-mesh topology (Claim 7). Edge tunnel end points within a domain are connected to all domain tunnel end points by a tunnel (Description, FIG. 1).

Hypothetical Combination of Prior Art for Obviousness:

Given the state of the art in network virtualization, especially with VXLAN and NVGRE being "most representative" prior to this patent's filing, a person of ordinary skill in the art would be familiar with the concepts of:

• Tunneling: Encapsulating network packets within another packet to traverse an underlying network, typically using L3 tunnels over L3 networks for L2 virtualization (Description, "overlay-based virtual network").
• Tunnel End Points (TEPs): Devices or software components responsible for encapsulating and de-encapsulating tunnel packets.
• Virtual Switching Instances (VSIs): Logical switches that perform L2 forwarding functions in a virtualized environment.
• Full-mesh topology: A network arrangement where every node is directly connected to every other node.
• L2 switching: Forwarding Ethernet frames based on MAC addresses.
• Virtual Network Identifiers (VNIDs): Used to distinguish different virtual networks.

Problem to be Solved: The patent explicitly states the problem: "restricts expandability, thereby causing a problem of restricting the size of the network (number of network nodes) where a VXLAN or NVGRE may be applied." This "long-felt need" for greater scalability in virtual networks would provide a strong motivation for a person of ordinary skill in the art to seek solutions.

Motivation to Combine Existing Concepts:

A person of ordinary skill in the art, facing the scalability limitations of VXLAN/NVGRE in a single, large full-mesh domain (where each TEP connects to every other TEP), would naturally look for ways to reduce the number of direct tunnel connections. Network architects commonly use hierarchical or federated approaches to scale networks that initially rely on flat or full-mesh designs.

Hypothetical Prior Art Combination:

1. Prior Art Reference A (e.g., describing VXLAN or NVGRE): This reference would teach the basic principles of overlay-based L2 network virtualization using L3 tunnels between TEPs in a full-mesh topology, and the use of VNIDs and L2 switching within a VSI (as explicitly mentioned in the background of US10020961). This reference would also disclose the problem of N*(N-1) tunnels limiting scalability.
2. Prior Art Reference B (e.g., describing hierarchical network design or gateway concepts): This reference would teach the general concept of using intermediate "gateway" or "border" nodes to connect smaller, isolated network segments to form a larger, more scalable network. This is a fundamental concept in networking to avoid full-mesh connections in large-scale systems. For example, in traditional IP routing, routers connect separate broadcast domains or subnets. In the context of virtual networks, this concept could be applied to virtual domains.
3. Prior Art Reference C (e.g., describing tunnel switching or relaying in a different context): This reference could disclose the idea of a network node that terminates a tunnel and then re-encapsulates the inner packet into a new tunnel for forwarding to a further destination, effectively acting as a "tunnel switch" or relay.

Motivation for Combination:

• Addressing the Scalability Problem: The primary motivation would be to overcome the N*(N-1) tunnel scaling issue identified in Reference A. A person of ordinary skill would recognize that introducing an intermediate layer of "gateway" or "domain" entities (from Reference B) could reduce the number of direct tunnels required between individual edge TEPs.
• Applying Hierarchical Design to Virtual Networks: The known principle of hierarchical network design (from Reference B) provides a clear motivation to divide a large virtual network into smaller, manageable "domains" and connect these domains via dedicated "domain" entities.
• Leveraging Tunnel Switching/Relaying: To enable communication between edge TEPs in different domains that are not directly connected, the "domain" entities would need to perform a "tunnel switching" or "relaying" function (as disclosed in Reference C). That is, they would receive a tunnel packet, de-encapsulate it to expose the inner L2 packet, perform L2 switching, and then re-encapsulate it in a new tunnel to the appropriate destination domain tunnel end point or edge tunnel end point. The patent explicitly describes the domain tunnel end point performing like a "tunnel switch" (Description, "an edge tunnel end point...logically operate as if it is connected to all the edge tunnel end points belonging to other domains with a tunnel in full-mesh topology").
• Maintaining L2 Semantics Across Domains: The goal of network virtualization, as taught by Reference A, is to create a virtual L2 network. Therefore, the intermediate "domain tunnel end points" would need to perform L2 switching (as performed by VSIs in Reference A) to maintain the L2 semantics across the interconnected domains. The modified flooding rules (Claim 2, 8) would be a logical refinement to prevent forwarding loops in such a multi-domain L2 network.

How the Combination Renders Claims Obvious:

By combining these concepts, a person of ordinary skill could arrive at the claimed invention:

• Tunnel Manager (Claims 1, 7, 13): The function of collecting tunnel end point information and connecting tunnels is a standard management task in any virtualized network (taught by Reference A). Extending this to manage both edge TEPs within a domain and domain TEPs between domains (as necessitated by the hierarchical design of Reference B) would be an obvious application.
• Tunnel Packet End Point & Tunnel Packet Generator (Claims 1, 7, 13): These components perform the standard tunneling functions (encapsulation/de-encapsulation) taught by Reference A, but applied in a multi-hop scenario where domain TEPs act as intermediate tunnel end points. The L3 tunnel header changing from (b) to (c) in FIG. 2 while passing through a domain tunnel end point (Description, "L3 tunnel header of the packet may be changed from (b) of FIG. 2 to (c) of FIG. 2 while going through the domain tunnel end point") exemplifies this known tunnel relaying/switching function.
• Domain VSI and L2 Switching (Claims 1, 7, 13): Integrating an L2 switching instance (VSI from Reference A) at the domain tunnel end point to forward the inner L2 packet after de-encapsulation, and before re-encapsulation, would be obvious to maintain the L2 virtual network across domains, consistent with Reference A and the relaying function of Reference C. The specific modified flooding rules (Claims 2, 8, and FIG. 5, 6) for domain VSIs, where flooding is restricted to either domain virtual ports (if input is edge VP) or edge virtual ports (if input is domain VP), would be an obvious engineering solution to prevent loops and optimize traffic within a hierarchical L2 network (a common problem in bridged networks, addressed by spanning tree protocols, etc., which a person of ordinary skill would be aware of).
• Multi-domain structure with full-mesh domain TEPs (Claim 7): This is the core structural innovation. Dividing a large virtual network into smaller domains, each with a "domain tunnel end point" (a gateway/border node from Reference B), and connecting these domain tunnel end points in a full-mesh (a known, albeit limited, connectivity pattern from Reference A, now applied at a higher level) directly addresses the scalability issue of Reference A by reducing the overall number of tunnels for edge devices.

In summary, the claims of US10020961, particularly concerning the multi-domain network virtualization approach with domain tunnel end points acting as tunnel switches, appear to be an obvious combination of known networking principles (tunneling, L2 switching, hierarchical network design, and tunnel relaying) in light of the recognized scalability problems of existing overlay networks like VXLAN and NVGRE. A person having ordinary skill in the art would have been motivated to combine these elements to improve the expandability of virtual L2 networks.