IP Library › Granted Patent US 12,581,318
Granted Patent B2
US 12,581,318 · App. 18/313,846 · Granted Mar 17, 2026

Network topology for efficient and performant virtual WANs

Inventors: Rachee Singh (Ithaca, NY); Anjali (Madison, WI)
Assignee: Microsoft Technology Licensing, LLC
H04W16/18H04L43/0852H04W24/08
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,581,318
App. No.
18/313,846
Granted
Mar 17, 2026
Kind
B2
Abstract

The present application relates to a system, apparatus, and method of selecting a topology for a virtual wide area network (vWAN). A wide area network (WAN) includes a plurality of geographically distributed entry points and a plurality of edge datacenters. Each edge datacenter is associated with an entry point and includes computing resources for hosting a vWAN hub. A vWAN manager is configured to: obtain latency measurements from different metropolitan regions to the entry points; receive a set of the metropolitan regions and a number of expected connections for each metropolitan region; and select a plurality of vWAN hub locations at selected edge datacenters, each selected edge datacenter being associated with a non-empty subset of the metropolitan regions. A weighted latency for the plurality of vWAN hub locations based on the latency measurements and the expected connections is minimized for a number of the selected edge datacenters.

Claims (45)

1 . A wireless wide area network (WAN), comprising:

a plurality of geographically distributed entry points including routers connected to other networks;

a plurality of edge datacenters, each edge datacenter associated with one of the plurality of geographically distributed entry points, each edge datacenter including computing resources capable of hosting a virtual WAN hub; and

a management datacenter including a memory storing computer-executable instructions, and at least one processor configured to execute the computer-executable instructions to:

obtain latency measurements from a plurality of clients, each client associated with one of a plurality of different metropolitan regions, to the plurality of geographically distributed entry points of the WAN;

receive, from an enterprise client, a set of the metropolitan regions and a number of expected connections for each metropolitan region; and

instantiate a plurality of virtual WAN hubs at selected edge datacenters of the plurality of edge datacenters, each selected edge datacenter being associated with a non-empty subset of the metropolitan regions, wherein a weighted latency for the selected edge datacenters based on the latency measurements and the expected connections is minimized for a number of the selected edge datacenters.

2 . The WAN of claim 1 , wherein to obtain the latency measurements, the at least one processor is configured to execute the computer-executable instructions to:

measure, for each of a plurality of clients, a latency from a respective client to an entry point of the WAN for a service within the WAN;

determine a metropolitan region of the respective client and the entry point for each latency measurement; and

aggregate the latency measurements for clients within metropolitan regions for each entry point having a threshold number of latency measurements to determine the latency measurement between a metropolitan region and an entry point.

3 . The WAN of claim 1 , wherein the at least one processor is configured to execute the computer-executable instructions to receive, from the enterprise client, a service level objective defining an expected latency for one or more of the metropolitan regions, wherein to instantiate the plurality of virtual WAN hubs, the at least one processor is configured to execute the computer-executable instructions to select only edge datacenters associated with entry points that satisfy the expected latency for a metropolitan region.

4 . The WAN of claim 1 , wherein to instantiate the plurality of virtual WAN hubs, the at least one processor is configured to execute the computer-executable instructions to solve a mixed integer linear program (MILP) to minimize a sum of latencies for selected entry points across the set of metropolitan regions for the number of expected connections for each metropolitan region.

5 . The WAN of claim 4 , wherein the MILP is subject to:

a first constraint that connections from one client metropolitan region go to only one virtual WAN hub associated with an entry point; and

a second constraint that only one edge datacenter associated with an entry point is selected for one metropolitan region.

6 . The WAN of claim 5 , wherein the MILP is further subject to a third constraint that the number of selected edge datacenters is less than a threshold.

7 . The WAN of claim 6 , wherein the at least one processor is configured to execute the computer-executable instructions to solve the MILP for different values of the threshold to determine points on a Pareto optimal frontier for the latency and the number of selected edge datacenters.

8 . The WAN of claim 7 , wherein the at least one processor is configured to execute the computer-executable instructions to select the number of edge datacenters based on a mean of a minimum feasible number of edge datacenters and a lowest latency number of edge datacenters.

9 . The WAN of claim 4 , wherein the at least one processor is configured to execute the computer-executable instructions to select a number of virtual WAN hubs to instantiate at each of the selected datacenters based on the number of expected connections for each metropolitan region associated with a respective edge datacenter and a maximum number of connections per virtual WAN hub.

10 . The WAN of claim 1 , wherein the at least one processor is configured to execute the computer-executable instructions to:

evaluate an average latency for the virtual WAN over a plurality of time windows using an exponentially weighted moving average;

reselect the selected datacenters based on the average latency; and

migrate one or more virtual WAN hubs to a newly selected edge datacenter from a datacenter associated with an edge datacenter that is no longer selected.

11 . A method comprising:

obtaining latency measurements from a plurality of clients, each client associated with one of a plurality of different metropolitan regions, to different entry points of a wide area network (WAN);

receiving, from an enterprise client, a set of the metropolitan regions and a number of expected connections for each metropolitan region; and

instantiating a plurality of virtual WAN hubs at selected datacenters within the WAN, each selected datacenter being associated with an entry point of the WAN and a non-empty subset of the metropolitan regions, wherein a weighted latency for the selected datacenters based on the latency measurements and the expected connections is on a Pareto optimal frontier of the weighted latency for a number of the selected datacenters.

12 . The method of claim 11 , wherein obtaining the latency measurements comprises:

measuring, for each of a plurality of clients, a latency from a respective client to an entry point of the WAN for a service within the WAN;

determining a metropolitan region of the respective client and the entry point for each latency measurement; and

aggregating the latency measurements for clients within metropolitan regions for each entry point having a threshold number of latency measurements to determine the latency measurement between a metropolitan region and an entry point.

13 . The method of claim 11 , further comprising receiving, from the enterprise client, a service level objective defining an expected latency for one or more of the metropolitan regions, wherein instantiating the plurality of virtual WAN hubs comprises selecting only datacenters associated with entry points that satisfy the expected latency for a metropolitan region.

14 . The method of claim 11 , wherein instantiating the plurality of virtual WAN hubs comprises solving a mixed integer linear program (MILP) to minimize a sum of latencies for selected entry points across the set of metropolitan regions for the number of expected connections for each metropolitan region.

15 . The method of claim 14 , wherein the MILP is subject to:

a first constraint that connections from one client metropolitan region go to only one virtual WAN hub associated with an entry point; and

a second constraint that only one datacenter associated with an entry point is selected for one client metropolitan.

16 . The method of claim 15 ; wherein the MILP is further subject to a third constraint that the number of entry points associated with a hub is less than a threshold.

17 . The method of claim 16 , wherein instantiating the plurality of virtual WAN hubs comprises solving the MILP for different values of the threshold to determine points on a Pareto optimal frontier.

18 . The method of claim 17 , wherein instantiating the plurality of virtual WAN hubs comprises selecting a number of entry points based on a mean of a minimum feasible number of entry points and a lowest latency number of entry points.

19 . The method of claim 14 , further comprising selecting a number of virtual WAN hubs to instantiate at each of the selected datacenters based on the number of expected connections for each metropolitan region associated with an entry point and a maximum number of connections per virtual WAN hub.

20 . The method of claim 11 , further comprising:

evaluating an average latency over a plurality of time windows using an exponentially weighted moving average;

reselecting the selected datacenters based on the average latency; and

migrating one or more virtual WAN hubs to a datacenter associated with a newly selected entry point from a datacenter associated with an entry point that is no longer selected.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 18, 2026
From: SINGH, RACHEE; ANJALI, _
To: MICROSOFT TECHNOLOGY LICENSING, LLC
Reel/Frame 073824/0122 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jul 28, 2023
From: SINGH, RACHEE
To: MICROSOFT TECHNOLOGY LICENSING, LLC
Reel/Frame 064415/0177 →
Continuity (2)
Provisional Application 63451850 · Mar 13, 2023
Related Publication 20240314577A1 · Sep 19, 2024
References Cited (40)
US 20200028746A1 · Zawadzki · 2020 [cited by examiner]
US 20220006756A1 · Ramaswamy · 2022 [cited by examiner]
US 20220201673A1 · Pandey · 2022 [cited by applicant]
US 20230379210A1 · Janakiraman · 2023 [cited by examiner]
US 20240414520A1 · Lu · 2024 [cited by examiner]
US 20250168101A1 · Ramaswamy · 2025 [cited by examiner]
CN 113196723A · 2021 [cited by applicant]
Singh et al., Cost-effective Cloud Edge Traffic Engineering with Cascara, 18th USENIX Symposium on Networked Systems Design and Implementation (NSDI 21), pp. 201-216, 2021 (Year: 2021). [cited by examiner]
Anjali et al., Cost-effective and performant virtual WANs with Cornifer, arXiv retrieved from https://arxiv.org/abs/2401.09620, 2024 (Year: 2024). [cited by examiner]
Singh et al., Cost-effective capacity provisioning in wide area networks with Shoofly, ACM SIGCOMM 2021, Aug. 2021 (Year: 2021). [cited by examiner]
Babasanmi, et al., “Measuring Cloud Latency in Africa,” 2022 IEEE 11th International Conference on Cloud Networking, Nov. 7, 2022, pp. 61-66. [cited by applicant]
International Search Report and Written Opinion received for PCT Application No. PCT/US2024/018452, Jun. 13, 2024, 16 pages. [cited by applicant]
Yan, et al., WAN as a service for cloud via software-defined network and open APIs, 2015 IEEE Conference on Computer Communications Workshops, Apr. 26, 2015, pp. 9-10. [cited by applicant]
“Amazon ec2: Secure and resizable compute capacity for virtually any workload”, Retrieved From: https://aws.amazon.com/ec2/, Retrieved on: Nov. 23, 2022, 7 Pages. [cited by applicant]
“AWS Cloud WAN”, Retrieved From: https://aws.amazon.com/cloud-wan/, Retrieved on: Nov. 23, 2022, 6 Pages. [cited by applicant]
“Introduction to IWAN and PfRv3”, Retrieved From: https://www.cisco.com/c/en/us/support/docs/ios-nx-os-software/performance-routing-pfr/200281-Introduction-To-IWAN-And-PfRv3.html, Feb. 19, 2021, 5 Pages. [cited by applicant]
“Magic WAN”, Retrieved From: https://www.cloudflare.com/magic-wan/, Retrieved on: Nov. 23, 2022, 13 Pages. [cited by applicant]
“scipy.stats.ttest_ind”, Retrieved From: https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.ttest_ind.html, Retrieved on: Nov. 23, 2022, 4 Pages. [cited by applicant]
“Virtual WAN”, Retrieved From: https://azure.microsoft.com/en-us/products/virtual-wan/#overview, Retrieved on: Nov. 23, 2022, 15 Pages. [cited by applicant]
“Virtual WAN pricing”, Retrieved From: https://azure.microsoft.com/en-us/pricing/details/virtual-wan/#pricing, Retrieved on: Nov. 23, 2022, 9 Pages. [cited by applicant]
Agrawal, et al., “A rewriting system for convex optimization problems”, In Journal of Control and Decision, vol. 5, Issue 1, Jan. 2, 2018, pp. 42-60. [cited by applicant]
Andersen, et al., “Resilient overlay networks”, In Proceedings of the eighteenth ACM symposium on Operating Systems Principles, Oct. 21, 2001, pp. 131-145. [cited by applicant]
Yap, “Taking the Edge off With Espresso: Scale, Reliability and Programmability for Global Internet Peering”, In Proceedings of the Conference of the ACM Special Interest Group on Data Communication, Aug. 7, 2017, pp. 4… [cited by applicant]
Balakrishnan, et al., “Analyzing stability in wide-area network performance”, In Proceedings of Analyzing stability In wide-area network performance, Jun. 1, 1997, pp. 2-12. [cited by applicant]
Calder, et al., “Analyzing the Performance of an Anycast CDN”, In Proceedings of the Internet Measurement Conference, Oct. 28, 2015, pp. 531-537. [cited by applicant]
Calder, et al., “Odin: Microsoft's Scalable Fault-Tolerant CDN Measurement system”, In Proceedings of 15th USENIX Symposium on Networked Systems Design and Implementation, Apr. 9, 2018, pp. 501-517. [cited by applicant]
Clark, et al., “Overlay Networks and the Future of the Internet”, In Proceedings of Communications and Strategies 63, Jul. 2006, pp. 109-129. [cited by applicant]
Diamond, et al., “CVXPY: A Python-Embedded Modeling Language for Convex Optimization”, In Journal of Machine Learning Research, vol. 17, Issue 83, Mar. 3, 2016, pp. 1-5. [cited by applicant]
Han, et al., “Topology Aware Overlay Networks”, In Proceedings of IEEE 24th Annual Joint Conference of the IEEE Computer and Communications Societies, vol. 4, Aug. 22, 2005, pp. 2554-2565. [cited by applicant]
Hong, et al., “Achieving High Utilization with Software-Driven WAN”, In Proceedings of the ACM SIGCOMM 2013 Conference on SIGCOMM, Aug. 27, 2013, pp. 15-26. [cited by applicant]
Jain, et al., “B4: Experience with a Globally-Deployed Software Defined WAN”, In Journal of ACM SIGCOMM Computer Communication Review, vol. 43, Issue 4, Aug. 27, 2013, pp. 3-14. [cited by applicant]
Jalaparti, et al., “Dynamic Pricing and Traffic Engineering for Timely Inter-Datacenter Transfers”, In Proceedings of the ACM SIGCOMM Conference, Aug. 22, 2016, pp. 73-86. [cited by applicant]
McGuire, et al., “Virtual WAN FAQ”, Retrieved From: https://learn.microsoft.com/en-us/azure/virtual-wan/virtual-wan-faq, Nov. 9, 2022, 23 Pages. [cited by applicant]
Pathan, et al., “Overlay networks: An akamai perspective”, In Publication of Advanced Content Delivery, Streaming, and Cloud Services, vol. 51, Issue 4, Oct. 2014, pp. 305-328. [cited by applicant]
Schlinker, et al., “Engineering Egress with Edge Fabric: Steering Oceans of Content to the World”, In Proceedings of the Conference of the ACM Special Interest Group on Data Communication, Aug. 7, 2017, pp. 418-431. [cited by applicant]
Sen, et al., “Can They Hear Me Now? A Case for a Client-Assisted Approach to Monitoring Wide-Area Wireless Networks”, In Proceedings of the ACM SIGCOMM conference on Internet measurement conference, Nov. 2, 2011, pp. 99… [cited by applicant]
Singh, et al., “Cost-Effective Capacity Provisioning in Wide Area Networks with Shoofly”, In Proceedings of the ACM SIGCOMM Conference, Aug. 9, 2021, pp. 534-546. [cited by applicant]
Singh, et al., “Cost-Effective Cloud Edge Traffic Engineering with Cascara”, In Proceedings of 18th USENIX Symposium on Networked Systems Design and Implementation (NSDI 21), Apr. 12, 2021, pp. 201-216. [cited by applicant]
Singh, et al., “Radwan: Rate Adaptive Wide Area Network”, In Proceedings of the Conference of the ACM Special Interest Group on Data Communication, Aug. 7, 2018, pp. 547-560. [cited by applicant]
International Preliminary Report on Patentability (Chapter I) received for PCT Application No. PCT/US2024/018452, mailed on Sep. 25, 2025, 10 Pages. [cited by applicant]