| NSF Org: |
CNS Division Of Computer and Network Systems |
| Recipient: |
|
| Initial Amendment Date: | July 3, 2017 |
| Latest Amendment Date: | March 11, 2020 |
| Award Number: | 1719165 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Darleen Fisher
CNS Division Of Computer and Network Systems CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | August 1, 2017 |
| End Date: | October 31, 2021 (Estimated) |
| Total Intended Award Amount: | $380,635.00 |
| Total Awarded Amount to Date: | $412,635.00 |
| Funds Obligated to Date: |
FY 2018 = $16,000.00 FY 2020 = $16,000.00 |
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
1776 E 13TH AVE EUGENE OR US 97403-1905 (541)346-5131 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
OR US 97403-5219 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): |
Special Projects - CNS, Networking Technology and Syst |
| Primary Program Source: |
01001819DB NSF RESEARCH & RELATED ACTIVIT 01002021DB NSF RESEARCH & RELATED ACTIVIT |
| Program Reference Code(s): |
|
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
Maps of the interconnections between pairs of networks or Autonomous Systems (ASes), such as Internet providers, are important to a wide range of parties, such as network researchers and network operators, whose activities depend on an accurate view of the Internet and its interconnections. Systematically mapping these interconnections is known to be a notoriously difficult problem. Here, the term 'mapping' refers to first inferring the presence and type of interconnections (e.g., public vs. private peering) and then geolocating them to the colocation facility where the interconnections are provided and utilized. Recently, the introduction of and growing demand for "virtual private interconnections" (VPIs) has been particularly disruptive for any such mapping efforts. VPIs are a new interconnection service that is offered at newly emerging infrastructures called "cloud exchanges" that some of the major colocation facility providers (e.g., Equinix and CoreSite) have started to operate in their major markets. By offering VPIs, these cloud exchanges enable traditional players (e.g., networks operating with an AS number) as well as non-traditional players (e.g., enterprises that do not own an AS number) to connect to any cloud provider that is present at the exchange, thereby enabling cloud-service-related traffic to bypass the public Internet. Moreover, in contrast to the traditional interconnection options, VPIs can be provisioned in real-time via the use of web-based portals. The ability of a large pool of non-traditional Internet players to establish interconnections in the form of VPIs and the increasingly dynamic nature of this type of interconnection have made the systematic mapping of today's Internet interconnections even more challenging. Not only are the existing techniques for mapping Internet interconnections incapable of detecting VPIs, but even when applied to the traditional types of interconnections, their applicability and accuracy remain largely unknown.
As more networks utilize cloud services, the concurrent use of traditional interconnections and cloud-exchange-provided VPIs is becoming more common and requires a new approach for discovering this 'hybrid' connectivity, studying its dynamics, and providing a first-of-its-kind assessment of the type of 'service-to-service' connectivity that is fueled by the 'cloudification' of today's Internet. The main goal of this research project is to develop such a new approach and design and rigorously evaluate the resulting suite of new techniques and open-source tools for inferring and geolocating interconnections in general, and VPIs at both cloud exchanges and certain IXPs (Internet eXchange Points), in particular. An online portal that offers useful visualization capabilities and provides access to a large archive of geo-aware interconnection maps will be developed as part of this project. The project will also enable other research groups to contribute to this important mapping effort, encourage the cross-validation of results by different research groups, and ultimately produce more accurate maps of the Internet's interconnections that can be expected to offer unique insights into new ways of detecting potential Internet vulnerabilities and performance bottlenecks. The project is complemented by an education and outreach plan that includes (i) the development of a curriculum for a new course and graduate level seminars, (ii) the participation of Masters level and undergraduate students in the proposed work (e.g., measurement, visualization), and (iii) the organization of a K-12 outreach activity where CS students volunteer to teach basic programming concepts to students at local elementary, middle, and high schools as part of after-school programs.
PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH
Note:
When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external
site maintained by the publisher. Some full text articles may not yet be available without a
charge during the embargo (administrative interval).
Some links on this page may take you to non-federal websites. Their policies may differ from
this site.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
The main goals of this research project are (i) to design, develop and rigorously evaluate new techniques for inferring and geolocating non-public-facing private portion of Internet topology (e.g. interconnection) in general, and "virtual private interconnections" (VPIs) at both cloud exchanges and certain IXPs (Internet eXchange Points), in particular, and (ii) to examine the implications of these interconnections at different granularities of observation (e.g., within a facility, between a cloud provider and rest of the Internet, multiple cloud providers, etc.). To this end, this project consists of the following four studies:
First, we develop mi2 , a new method for identifying and geolocating (i.e. mapping) Internet interconnections inside a given colocation facility. mi2 infers the existence of interconnections from localized traceroutes and use the Belief Propagation algorithm on a specially defined Markov Random Field graphical model to geolocate them to a target colocation facility. We evaluate mi2 by applying it initially to a small set of US-based colocation facilities and compare our its results against two recently developed related techniques. Our measurements identify new infrastructures in the form of “cloud exchanges” that offer new types of interconnections called “virtual private interconnections”.
Second, we conduct a third-party measurement study to reveal all the peerings between Amazon and the rest of the Internet. We develop a new technique for inferring these peering links and pay special attention to inferring the "virtual private interconnections" (VPIs) associated with this largest cloud provider. We also present and evaluate a new method for geo-locating each end of the inferred interconnections or peering links. Our study provides a first look at Amazon’s peering fabric. In particular, by grouping Amazon’s peerings based on their key features, we illustrate the specific role that each group of peerings plays in how Amazon peers with other networks.
Third, we conduct a cloud-centric measurement study of a coast-to-coast multi-cloud deployment that a typical modern enterprise located in the US may adopt. We deploy VMs in two regions (i.e., VA and CA) of each one of three large cloud providers (i.e., AWS, Azure, and GCP) and connect them using three different connectivity options: (i) transit provider-based best-effort public Internet (BEP), (ii) third-party provider-based private (TPP) connectivity, and (iii) Cloud provider-based private (CPP) connectivity. Our active measurements in this real-world multi-cloud deployment provide new insights into variability in the performance of TPP, the stability in performance and topology of CPP, and the absence of transit providers for CPP.
Fourth, we design and develop Tondbaz, a multi-cloud overlay service that empowers enterprises with tools necessary to measure and manage their overlays in a performance-aware and cost-effective manner. In demonstrating the efficacy of Tondbaz, we first present a third-party, cloud-centric measurement study to understand and examine the delay and traffic-cost profiles of cloud backbones by deploying virtual machines in three global-scale cloud providers (i.e. AWS, Azure, and GCP). Our measurements reveal new insights such as a best-in-class backbone of AWS and a general lack of path and delay asymmetries in cloud backbones. Next, we revisit the classical problem of overlay construction in a multi-cloud setting. By framing the problem as a simple optimization formulation and demonstrating its solution in practical deployments, we show that Tondbaz can achieve varying amounts of reduction in delay (e.g., up to 3x) for several national and international multi-cloud paths.
We have publicly shared some of our datasets to enable cross examinations of our results by other researchers. Our results were disseminated through project web pages and published in major journals (ToN) and conferences (IMC, PAM). We have also developed and offer an undergraduate course and a few graduate seminars in Internet measurement. This project offered training to two Ph.D. students and multiple M.S. and undergraduate students.
Last Modified: 05/10/2022
Modified by: Reza Rejaie
Please report errors in award information by writing to: awardsearch@nsf.gov.
