1. Characterizing the Two-Tier Gnutella Topology  Gnutella, FastTrack, and eDonkey use two-tier overlay topologies.  Our initial study focuses on Gnutella.  Top-level peers form the core overlay.  Each leaf connects to a few top-level peers. 5. Ongoing Work 2. Two-Tier Topologies  Characterizing file distribution and query workload  Characterizing Kademlia-based DHTs  Examination of performance bottlenecks in BitTorrent Top-to-Leaf Degree Distribution 4. Results Top-to-Top Degree Distribution Leaf-to-Top Degree Distribution Fig. 1 Daniel Stutzbach and Reza Rejaie – University of Oregon The Ion P2P Project: http://mirage.cs.uoregon.edu/P2P GraphPath Lengths Lengths of Random Clustering Coefficient CC of Random Modern Gnutella 4.17—4.233.750.0180.00038 Older Gnutella 3.30—4.423.660.020.002 Movie Actors3.652.990.790.00027 Power Grid18.712.40.080.005 Small World Properties Path Length Distribution  Peer degree is fairly homogeneous, not power-law.  Most top-level peers have a degree around 30 (Fig. 1).  Under 30, degree is nearly uniformly distributed (Fig. 1).  A power-law was reported by previous studies.  Our prior work shows that slow crawling can erroneously lead to a power-law degree distribution (Fig 2).  Degree distributions between tiers are also fairly homogenous.  The number of leaves per top-level peer is similar. Version differences cause different spikes (Fig. 3).  Most leaf peers have a very low degree, but a small number have a high degree (Fig. 4)  Despite exponential growth in size, overlay path lengths in Gnutella are very short (Fig. 5 and Fig. 6).  60% of top-level paths are exactly four hops in length.  99.5% of top-level paths are five hops or less.  Leaf-to-leaf paths are 1 or 2 hops longer, on average.  Gnutella is not power-law, but is still a small world.  Path lengths are close to same-size random graphs.  The top-level overlay is not tightly clustered (0.018).  However, it is much more clustered than same-size random graphs (0.018 >> 0.00038).  Characterizing & modeling the dynamics of overlay topologies: 1) Peer churn, 2) Edge churn  Developing an overlay topology generator for simulation 3. Approach 1. Motivation  Most of the large file sharing Peer-to-Peer (P2P) applications with millions of users are based on unstructured, two-tier overlay topologies.  Characterizing the unstructured overlays of these applications is important for design and evaluation.  Characterizing P2P overlays requires capturing accurate and fine-grained snapshots of the overlays.  Snapshots (as graphs) are captured with a crawler, recording peers (as nodes) & connections (as edges).  Captured snapshots by a slow crawler can be distorted due to 1) dynamic changes in the overlay during a crawl, 2) peers unreachable by the crawler.  Previous studies are outdated, used slow crawlers (1 or 2 hours), and did not examine the accuracy of their captured snapshots.  We developed a parallel and tunable crawler, Cruiser.  Cruiser increases crawling speed by  Using a master-slave architecture  Crawling many peers in parallel  Leveraging the two-tier topology  Cruiser captures an accurate Gnutella snapshot with 1-million peers in around 7 minutes (140k peers/min). Inaccuracy of Slow Crawling Fig. 5 Fig. 6 Fig. 4 Fig. 2 Fig. 3