In particular, I would like to thank Prof. Rafael Saavedra, Gene Tsudik, Abhjit Khale, Lee Breslau, Steve Hotz, Doug Fang, Danny Mitzel, Ron Cocchi, Sugih Jamin, Shih-Hao Li, Jong-Suk Ahn, Daniel Zappala, Brenda Tim m erm an, John Noll, Kraig Meyer, Louie Ramos and Ari Medvinski.
it manages a Take the Internet news distribution system [31] as an example: highly dynamic, weakly consistent, gigabyte database replicated at thousands of autonomous adm inistrative domains, yet responds to queries in seconds.
1.2 What Current Algorithms Lack As existing naming services and distributed file systems have dem onstrated, the problem of replicating data th at can be partitioned into autonom ously managed subspaces has well-known solutions.
1.5 Dissertation Overview and Outline In the previous sections, we argued th at efficient replication algorithms keep replicas weakly consistent by flooding updates between them and don’t necessarily have to rely on existing m ulticast transport protocols.
This is because hidalgo, which is one of the machines th a t has the biggest memory in the USC Networking and D istributed Systems Laboratory, is usually running big sim ulations, and had the additional load of being the group m aster during this experim ent.