Hobbit clusters are the groups of IPv4 addresses that are topologically proximate. Hobbit clusters can be useful for reducing measurement loads, generating representative samples and characterizing the addresses of the same type. Hobbit clusters are publicly available.
Hobbit inspects the last-hop routers of addresses to determine topological proximity. Hobbit determines addresses to be topologically proximate if they have identical last-hop routers or have different last-hop routers due to load-balancing (rather than typical routing). In order to distinguish between the differences in last-hop routers due to load-balancing and typical routing, Hobbit uses that typical routing is based on non-overlapping prefixes whereas load-balancing uses hashing. See our paper for more details about our technique and example applications.
Youndo Lee and Neil Spring.
Identifying and Aggregating Homogeneous IPv4 /24 Blocks with Hobbit.
In Proceedings of the ACM SIGCOMM Internet Measurement Conference (IMC), pages 151-165, Santa Monica, November 2016.
Each data file contains a single snapshot of Hobbit clusters taken from a single source at UMD.
Each line in the file consists of the prefixes of /24 blocks within the same Hobbit cluster separated by ",". For example, if /24 blocks 1.1.1.0/24, 2.2.2.0/24 and 3.3.3.0/24 are in the same Hobbit cluster, the file contains the line "1.1.1,2.2.2,3.3.3".
2016-03-23 (The snapshot used for analyses in the IMC 2016 paper.)
2016-11-02
(New snapshots will be added periodically.)