1. h t t p : / / w w w. e e c s. c a s e. e d u / DipZoom A Marketplace for Internet Measurements Faculty: Michael Rabinovich Students: Sipat Tiukose, Zhihua Wen Ack: Limin Wang EECS Department Case Western Reserve University

2. DipZoom 2 Internet Measurements A-priori measurement platforms (e.g. IDMaps)  Great for large-scale characterizations  Not focused enough for some applications On-demand measurements (Keynote, Scriptroute/PlanetLab, traceroute servers)  Hard to deploy sufficient platform to serve unpredictable needs  Vulnerable to being “gamed”  Limited choice of measurements

3. DipZoom 3 Example: Latency Measurements  Measuring latency between a laptop connected through VZACCESS BroadbandAccess and a Linux PC on Case network  RTT measured by average of 2000 pings: 280 msec King Measurements

4. DipZoom 4 DIPZoom Goals Fulfill the need for zoomed-in, on-demand measurements Provide infrastructure that scales with The Internet – Scaling with Internet size – Scaling with Internet diversity (devices, connectivity) – Scaling with measured metrics and regime Simplify staging of measurement experiments – Complex multi-step measurements – Distributed measurements across multiple measuring hosts (Hence the name: Deep Internet Performance Zoom)

5. DipZoom 5  Coax Internet users to become measurement providers  Deploy a matchmaking service instead of measurement infrastructure  Use market driven approach with real money as the means to control the system DIPZoom Approach

6. DipZoom 6 DipZoom: Deep Internet Performance Zoom-In  Anyone can offer measurements  Anyone can request measurements  Anyone can offer measuring software  Participants are free to set their prices, compete for requests, bid and solicit bids, etc.  Facilitates an open echosystem, “ebay for Internet measurements”  Client library allows programmatic access to the entire platform from one computer

7. DipZoom 7 Some Questions  Will anyone want to become a provider?  Seti@home, upromise.com, gomez.com suggest “yes”.  Will anyone pay for the measurements?  Keynote and Gomez suggest “yes”  Keynote is in business with 53M dollars revenue last year.

8. DipZoom 8 Related Work  Gomez.com  Closed commercial system  DIMES, Traceroute@home  No incentives  Users participate in a particular measurement experiment  Users can’t ask for a measurement  On-demand measurement systems (Keynote, Scriptroute, NIMI, GNP, King)  Hard to scale with Internet  Hard to stage experiments

9. DipZoom 9 System Overview

10. DipZoom 10 Issues  Security  Induced DoS attacks against measurement target  Open ports on measuring hosts  High-level DoS attack against measuring host  Unauthorized resources access  Payment trust  Trusted core helps  Replay-based cheating

11. DipZoom 11 Issues  Measurement trust  Fake Measuring Point (MP) registrations  Measuring Point (MP) impersonation  Fake measurements  Traversing firewalls and NATs

12. DipZoom 12 Core Needs  Integrity of measuring software  Globally unique ID of measuring host (MHID)  Duplicate detection + request/response matching  Measurement rate limiting Building blocks of a solution: – Unique embedded secrets – GUID or MAC address or Host ID – Nonces (“numbers used once”) – Rate limiting (per measurement host and per target) – Ranking and calibration

13. DipZoom 13 Ranking and Calibration  Security measures raise the bar but do not guarantee protection from malicious MH.  If can’t protect - detect and blacklist!  Deploy calibrating measurement targets  Purchase measurements from suspect MHs  Compare responses with passive measurements by calibrating hosts  Not all measurements can be calibrated (wget vs. ping vs. nslookup) Can calibrating hosts be gamed? –Keynote advertises its measuring hosts location –Calibrating hosts are secret –The risk of blacklisting deters data mining

14. DipZoom 14 Usage Example 1: Accuracy of King Measurements King (between well-connected) King (between well-connected and residential) King validation using DIPZoom King accuracy is sensitive to the type of access network Inaccuracy is biased to under-estimation

15. DipZoom 15 Usage Example 2: Accuracy of GNP measurements GNP is more accurate and less sensitive to the type of access connection The number of measurements needed to achieve accuracy depends on access connection 10 Measurements per pair 2.5 measurements per pair

16. DipZoom 16 Usage Example 3: Lowering the Barrier of Entry for Experimenters  Assessing the quality of server selection by Akamai  Discovering a large number of Akamai servers  Comparing the performance of Akamai-selected server with other Akamai servers  Currently:  Intricate experiments  Subject of research papers (Su et al. SIGCOMM’06; Krishnamurthy et al. IMW’01)  With DIPZoom:  1-month course project for undergrads  Done from students’ laptops with no special arrangements from measurement providers

17. DipZoom 17 Status First release available for download  http://dipzoom.case.edu/ http://dipzoom.case.edu/  Includes a GUI client and client class library for scripted experiments  Includes NAT/firewall traversal  Works on Windows, Linux, Mac OS X  Now offers ping, wget, traceroute, and nslookup.  No payments yet  No ranking and calibration  Around 150 MHs deployed on PlanetLab and several home computers

18. DipZoom 18 Next Up  Pluggable measurement tools  DipZoom veneer over existing infrastructures (e.g. Scriptroute)  Down the line:  Ranking and calibration  Payments  Distributed DipZoom core

19. DipZoom 19 Summary  Growing Internet diversity (devices, links, applications) entails growing needs for focused measurements  Proprietary platforms are insufficient  Measurements are increasingly complex  DipZoom: a facilitator instead of infrastructure  Open system (pricing, measurements, participants)  Unified programmatic access to entire platform  Market approach to system control  Many challenges ahead Web site for more information: http://dipzoom.case.edu/

20. h t t p : / / w w w. e e c s. c a s e. e d u / Questions and Comments