Skynet: A Cloud-Hopping Data Transfer Architecture Aleksandar Kuzmanovic

Slides:



Advertisements
Similar presentations
Network Aware Forward Caching Presenter: Alexandre Gerber Jeffrey Erman, Mohammad T. Hajiaghayi, Dan Pei, Oliver Spatscheck AT&T Labs Research April 24.
Advertisements

Yahoo! CDN and Peering Flávio Amaral – Br-Ops May 27th, 2008.
Multihoming and Multi-path Routing
1 IEEE Media Independent Handoff Overview of services and scenarios for 3GPP2 Stefano M. Faccin Liaison officer to 3GPP2.
Network Resource Broker for IPTV in Cloud Computing Lei Liang, Dan He University of Surrey, UK OGF 27, G2C Workshop 15 Oct 2009 Banff,
GridNets2009, Athens 08/09/09 Business Models, Accounting and Billing Concepts in Grid-aware Networks Serafim Kotrotsos 1, Peter Racz 2, Cristian Morariu.
Google-based Traffic Classification Aleksandar Kuzmanovic Northwestern University IEEE Computer Communications Workshop (CCW 08) October 23, 2008
Aditya Akella An Empirical Evaluation of Wide-Area Internet Bottlenecks Aditya Akella with Srinivasan Seshan and Anees Shaikh IMC 2003.
A Media Subscription Service By Peter Kriens CEO aQute OSGi Technology Officer
The Evolution of P2P Technology Robert Levitan, CEO Pando Networks P2P MEDIA SUMMIT, CES 2008.
When TCP Friendliness Becomes Harmful Amit Mondal Aleksandar Kuzmanovic Northwestern University
Video Services over Software-Defined Networks
Dynamic Replica Placement for Scalable Content Delivery Yan Chen, Randy H. Katz, John D. Kubiatowicz {yanchen, randy, EECS Department.
Internet Indirection Infrastructure (i3 ) Ion Stoica, Daniel Adkins, Shelley Zhuang, Scott Shenker, Sonesh Surana UC Berkeley SIGCOMM 2002 Presented by:
Justine Sherry*, Shaddi Hasan*, Colin Scott*, Arvind Krishnamurthy†,
1 Jin Li Microsoft Research. Outline The Upcoming Video Tidal Wave Internet Infrastructure: Data Center/CDN/P2P P2P in Microsoft Locality aware P2P Conclusions.
The Platform as a Service Model for Networking Eric Keller, Jennifer Rexford Princeton University INM/WREN 2010.
Employee & Manager Self Service Overview
Current thinking. Subject to change. October 2011 BENEFITS OF BUILDING AN IN-HOUSE CONTENT DISTRIBUTION NETWORK (CDN)
Taming User-Generated Content in Mobile Networks via Drop Zones Ionut Trestian Supranamaya Ranjan Aleksandar Kuzmanovic Antonio Nucci Northwestern University.
Countering DoS Attacks with Stateless Multipath Overlays Presented by Yan Zhang.
© Blue Coat Systems, Inc All Rights Reserved. APTs Are Not a New Type of Malware 1 Source: BC Labs Report: Advanced Persistent Threats.
IP Multicast Information management 2 Groep T Leuven – Information department 2/14 Agenda •Why IP Multicast ? •Multicast fundamentals •Intradomain.
Copyright © AIIM | All rights reserved. #AIIM The Global Community of Information Professionals aiim.org Information Management and Social Media Jesse.
Adding Up In Chunks.
25 seconds left…...
Subtraction: Adding UP
Bytindex – Data Optimization Concept Technology briefing
University of Minnesota Optimizing MapReduce Provisioning in the Cloud Michael Cardosa, Aameek Singh†, Himabindu Pucha†, Abhishek Chandra
Microsoft Volume Licensing
19 Historical overview Main challenge: How to distribute content in high quality over the Internet cost-effectively? • Traditional “Best-effort” model:
The Power of Explicit Congestion Notification Aleksandar Kuzmanovic Northwestern University
Rice Networks Group Aleksandar Kuzmanovic & Edward W. Knightly TCP-LP: A Distributed Algorithm for Low Priority Data Transfer.
Criticisms of I3 Jack Lange. General Issues ► Design ► Performance ► Practicality.
Content In The Cloud Conference 2012 International CES Entertainment Industry Benefits and Drawbacks of Cloud-Delivered Content: Innovation and Flexibility.
Drafting Behind Akamai (Travelocity-Based Detouring) Aleksandar Kuzmanovic Northwestern University Joint work with: A. Su, D. Choffnes, and F. Bustamante.
1 Drafting Behind Akamai (Travelocity-Based Detouring) AoJan Su, David R. Choffnes, Aleksandar Kuzmanovic, and Fabian E. Bustamante Department of Electrical.
1 Web Content Delivery Reading: Section and COS 461: Computer Networks Spring 2007 (MW 1:30-2:50 in Friend 004) Ioannis Avramopoulos Instructor:
Web Caching and CDNs March 3, Content Distribution Motivation –Network path from server to client is slow/congested –Web server is overloaded Web.
Presented by Sujit Tilak. Evolution of Client/Server Architecture Clients & Server on different computer systems Local Area Network for Server and Client.
P2P Games Conference “Attributes of the Gaming Cloud?” Norman Henderson ASANKYA
Content Distribution March 8, : Application Layer1.
1 One-Click Hosting Services: A File-Sharing Hideout Demetris Antoniades Evangelos P. Markatos ICS-FORTH Heraklion,
Towards a Safe Playground for HTTPS and Middle-Boxes with QoS2 Zhenyu Zhou CS Dept., Duke University.
1 MultimEDia transport for mobIlE Video AppLications 9 th Concertation Meeting Brussels, 13 th February 2012 MEDIEVAL Consortium.
Ao-Jan Su, David R. Choffnes, Fabián E. Bustamante and Aleksandar Kuzmanovic Department of EECS Northwestern University Relative Network Positioning via.
CS An Overlay Routing Scheme For Moving Large Files Su Zhang Kai Xu.
CDN: Content Distribution Networks  References:  CS613 textbook, “Computer Networking – A Top-Down Approach”, 6 th edition. Chapter  The text.
Economic Incentives in Information- Centric Networking: Implications for Protocol Design and Public Policy Group Members: Muhammad Kamran Siddique Adel.
Upload, Process, and Deliver Digital Media Assets from Everywhere and at Any Time with Reelway and the Microsoft Azure Cloud MICROSOFT ISV PROFILE: REELWAY.
A Survey on Network Storage Ning Zong Haibin Song Richard Alimi Richard Yang.
Skynet: A Cloud-Based Data Transfer Architecture Aleksandar Kuzmanovic
CS 6401 Overlay Networks Outline Overlay networks overview Routing overlays Resilient Overlay Networks Content Distribution Networks.
Overlay Networks : An Akamai Perspective
Flight is a SaaS Solution that Accelerates the Secure Transfer of Large Files and Data Sets Into and Out of Microsoft Azure Blob Storage MICROSOFT AZURE.
Peering Strategies for Operators Building critical mass of peers.
09/13/04 CDA 6506 Network Architecture and Client/Server Computing Peer-to-Peer Computing and Content Distribution Networks by Zornitza Genova Prodanoff.
An Analysis of Internet Content Delivery Systems 19 rd November, 2007 Youngsub CSE, SNU.
WHAT'S THE DIFFERENCE BETWEEN A WEB APPLICATION STREAMING NETWORK AND A CDN? INSTART LOGIC.
Drafting Behind Akamai (Travelocity-Based Detouring) Ao-Jan Su, David R. Choffnes, Aleksandar Kuzmanovic and Fabián E. Bustamante Department of EECS Northwestern.
Scalable Web Apps Target this solution to brand leaders responsible for customer engagement and roll-out of global marketing campaigns. Implement scenarios.
Caching Temporary storage of frequently accessed data (duplicating original data stored somewhere else) Reduces access time/latency for clients Reduces.
1. Public Network - Each Rackspace Cloud Server has two networks
Scalable Web Apps Target this solution to brand leaders responsible for customer engagement and roll-out of global marketing campaigns. Implement scenarios.
TCP-LP: A Distributed Algorithm for Low Priority Data Transfer
2017 Real Questions
ECE 671 – Lecture 16 Content Distribution Networks
The Internet and Electronic Mail
Fixing the Internet: Think Locally, Impact Globally
EE 122: Lecture 22 (Overlay Networks)
Presentation transcript:

Skynet: A Cloud-Hopping Data Transfer Architecture Aleksandar Kuzmanovic

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture The Rise of Hyper Giants Top 150 content providers account for 50% of traffic Top 30 content providers account for 35% of traffic 2

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Hyper Giants Morphing into CDNs 3

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Hyper Giants CDN Examples More than 60% of traffic flows directly between Google and consumer networks –Good for Google and consumer networks Facebook applies the same approach –More than 25% of Facebook runs through direct peerings with last- mile providers Microsoft started building its own CDN Google Global Cache Server 4

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Hyper Giants CDNs Properties Hyper giants CDNs are different from traditional CDNs (e.g., Akamais): –Support bi-directional data transfers, i.e., both uploads and downloads Needed for various Web 2.0 apps –Agile replication of content towards anticipated receivers Research question: –Can we build a generic data transfer architecture on top of clouds? 5

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Legacy Data Transfer 6

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Cloud-Based Data Transfer 7

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Why Should This Work? 8 –Cut the e2e Internet path into several shorter-RTT path chunks –Avoid Internet bottlenecks –Benefit from agile cloud replication source destination Cloud-based path

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Deployment Scenarios The bad –Free-riding not likely to be liked by clouds: Private resources used by 3 rd parties –The problem is a transparent data transfer service provided by clouds Hard to detect because: –Large amounts of data »we were able to push 100 Gbytes with no problems –Opening multiple accounts is straightforward The good –Potential for selling data acceleration services to 3 rd parties such that the performance of host applications is not degraded 9

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Cloud-Based Data Transfer Services 10 Multicast Multi-cloud hopping Multi-cloud homing

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Remaining Outline Cloud selection and properties Per-cloud data transfer performance Decomposing cloud-based data transfers Multi-cloud homing Cloud-based multicast Cloud path consistency Cloud-hopping overlays Cloud-supported overlays 11

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Investigated Carrier Applications Region # of IPs GmailHotmail utube lickr DUDUDUDU N. America Europe Asia S. America Oceania Unknowns Total

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Per-Cloud Performance Cloud-based paths can often, consistently and significantly outperform regular Internet paths 63.5% 42.3% Transcoding effects

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Per-Cloud Performance Cloud-based paths can often, consistently and significantly outperform regular Internet paths 41% 19.4% Fully transparent data transfer

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Transfer-Time Variance Variance of transfer times is much smaller for cloud-based transfers then for Internet paths Internet

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Remaining Outline Cloud selection and properties Per-cloud data transfer performance Decomposing cloud-based data transfers Multi-cloud homing Cloud-based multicast Cloud path consistency Cloud-hopping overlays Cloud-supported overlays 16

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Upload Latency Upload latency depends upon the scale and distribution of upload servers

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Replication Latency Different replication policies No replication Replication

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Download Latency Pushing data closer to end users pays off 14 sec 73 sec rate limiting

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Cloud-Based Multicast For multicast group sizes > 25, cloud-based multicast outperforms direct multicast for all clouds

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Remaining Outline Cloud selection and properties Per-cloud data transfer performance Decomposing cloud-based data transfers Multi-cloud homing Cloud-based multicast Cloud path consistency Cloud-hopping overlays Cloud-supported overlays 21

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Cloud-Supported Overlay Cloud-based paths bring additional benefits in overlay scenarios 35% improvement for files that experience longest transfer times

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture Conclusions The rise of hyper giants and their transparency enable cloud-based data transfers We demonstrated the feasibility of building data transfer services on top of clouds and their superior performance –Opens novel security challenges and may lead to an arms race in this area –Shows significant potentials for building explicit ties between legacy- and the new Internet 23

Aleksandar Kuzmanovic Skynet: A Cloud-Based Data Transfer Architecture 24 Thank You! Questions?