1 / 25

Overlay/P2P Networks

Overlay/P2P Networks. CS 641 Computer Networks. IP world. point-to-point communication unicast, multicast hierarchical addressing address related to physical end-point routing protocols RIP, OSPF, BGP …. Issues with IP. Rigid Infrastructure does/has not changed much from inception

svein
Download Presentation

Overlay/P2P Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Overlay/P2P Networks CS 641 Computer Networks Overlay Networks

  2. IP world • point-to-point communication • unicast, multicast • hierarchical addressing • address related to physical end-point • routing protocols • RIP, OSPF, BGP … Overlay Networks

  3. Issues with IP • Rigid • Infrastructure does/has not changed much from inception • E.g., multicast, QoS, routing algorithms … • High-barrier for innovation • Routing • Slow to detect/converge • Application-specific routing not possible • Audio app may want different delay from data app • Cannot adaptively change routing policy • Mobility • End-points change => IPs change=> Non-transparent! Overlay Networks

  4. Solution: Look upwards! • Application layer-based solutions • no changes required at IP substrate • lower-barrier for change/innovation • wasn’t IP supposed to provide support to app layer? • Overlay Network • Network on top of a network • Host (application) logical link to another host • Via multiple lower-level physical links • Extra functionality at application layer • Routing, Naming Overlay Networks

  5. Overlay Network View A E A -> B A -> D A -> C via B A-> D via B A -> C via B A-> C via D A-> E via D B C Router Host D physical link logical link Overlay Networks

  6. Overlay Networks • Communication Overlays • Optimize routing • RON (Resilient Overlay Networks) • Data Overlays • Data dissemination • Chord, CAN • Freenet, Napster, Gnutella … • Application-level multicast, anycast • multimedia, stocks, publish-subscribe … Overlay Networks

  7. How Robust is Internet Routing? • Slow outage detection and recovery • Inability to detect badly performing paths • Inability to efficiently leverage redundant paths • Inability to perform application-specific routing • Inability to express sophisticated routing policy Overlay Networks

  8. RON: Reliable Overlay Networks • Goals • Enable a group of nodes to communicate even in presence of problems on underlying IP paths • Integrate routing and path selection with applications • App-specific notions of performance • Apps define and react to failures • Framework for implementing expressive routing policies • Classify packets for rate control etc. • Limited path pubilicty Overlay Networks

  9. Reliability via path monitoring and re-routing Reliability via path monitoring and re-routing RON: Routing Using Overlays • Cooperating end-systems in different routing domains can conspire to do better than scalable wide-area protocols Scalable BGP-based IP routing substrate • Types of failures • Outages: Configuration/op errors, software errors, backholes, etc. • Performance failures: Severe congestion, DoS attacks, etc. Overlay Networks

  10. Link failures appear in many different places, regardless of duration Not the case that a few bad links Need to react to failures and not avoid known trouble links Can RON work? Overlay Networks

  11. Non-trivial number of outages last for greater than 5 minutes Several are multi-connected Have more than 2 links Can react, most likely alternate path might exist Can RON work? Overlay Networks

  12. RON Design • Hosts form a limited-size overlay • Hosts cooperate to forward data on behalf on any two nodes of overlay • Aggressively probe paths connecting overlay nodes • Quick failure detection and recovery • Path metrics • Latency, Packet loss, throughput • BGP: connectivity as long as BGP packets go through • Exchange path metrics to build forwarding tables • Link-state routing Overlay Networks

  13. Active Proving • RON probes every other node • PROBE_INTERVAL plus a random jitter of 1/3 PROBE_INTERVAL • A probe not returned in PROBE_TIMEOUT is considered loss Overlay Networks

  14. Outage Detection • On the loss of a probe, several consecutive probes spaced by PROBE_TIMEOUT are sent out • If OUTAGE_THRESH probes elicit no response the path is considered “dead” • If even one probe gets a response then high frequency probing is cancelled • Paths experiencing outages are rated on their packet loss history Overlay Networks

  15. Routing Metrics • Latency • lati = α x lati + (1- α) x new_samplei • latpath = summation of latencies of links on path • Loss Rate • Average loss rate over a window of k samples • Assuming losses on links are independent • lossratepath= 1 –Πi (1- lossratei) ( i є path) • Throughput • Function of loss and latency • p = one-way packet loss probability • rtt = e2e delay Overlay Networks

  16. Summarization mechanism Clients/Apps can leave/join, health always estimated Routing policies use information from database Smaller data units can be exchanged RON Performance Database Overlay Networks

  17. Experimental Evaluation • Raw measurement data consisted of probe packets • To probe each RON node independently repeated the following steps • Pick a random node j • Pick a probe-type from one of {direct, latency, loss} using round-robin. • Send probe to j • Delay for a random interval between 1 and 2 seconds Overlay Networks

  18. Experimental Evaluation • Two distinct datasets • RON1 • 64 hours between 3/21/2001 and 3/23/2001 • 12 nodes with 132 distinct paths • Traverses 36 different AS’s and 74 distinct inter-AS links • RON2 • 85 hours between 5/7/2001 and 5/11/2001 • 16 nodes with 240 distinct paths • Traverses 50 AS’s and 118 different AS links Overlay Networks

  19. RON greatly improves loss-rate 30-min average loss rate on Internet RON loss rate never more than 30% 13,000 samples 30-min average loss rate with RON Overlay Networks

  20. An order-of-magnitude fewer failures 30-minute average loss rates 6,825 “path hours” represented here 5 “path hours” of essentially complete outage 16 “path hours” of TCP outage RON routed around all of these! Overlay Networks

  21. Resilience against DoS attacks Overlay Networks

  22. Latency • RON improves slower paths • 11% saw improvements of 40 ms or more in RON1 • 8.2% saw improvements of 40 ms or more in RON2 Overlay Networks

  23. Throughput Improvement 132 RON1 paths Overlay Networks

  24. RON Summary • Demonstrates feasibility of overlay networks to improve performance/reliability • Good starting point as a testbed to build on • Tested with real deployments Overlay Networks

  25. RON Issues • Scalability • Works only with limited size of overlay? • What are exact overheads? • Not studied explicitly • What is tradeoff between reliability and probing rates? • What if several overlay exists? • Can reactive routing benefit? • Traffic carried on links multiple times • How will ISP relationships be effected, if serving others traffic? Overlay Networks

More Related