[Simple version of sample Fall 2007 15-712 outline.] Low-Overhead Byzantine Fault-Tolerant Storage James Hendricks et. al It is okay to post this outline on the web. 10 pages. Abstract -- Previous Byzantine fault-tolerant block storage protocols performed poorly because they lacked erasure-coding or requried expensive protocol techniques. Our protocol employs novel mechanisms to optimize for the common case and a new cryptographic primitive, homomorphic fingerprinting, to verify distributed erasure-coded data. 1. Introduction -- 1 page. Explains why Byzantine fault-tolerant storage is important and why prior solutions performed poorly. 2. Background -- 1.5 pages. Reviews redundancy techniques used for reliability, discusses erasure-coding for write performance, discusses costs of Byzantine fault-tolerant storage, discusses why prior solutions performed poorly (no erasure coding). 3. Design -- 2 pages. 4. Implementation -- 2 pages. Describes implementation of our protocol and three competing protocols (PASIS-emulation, crash-tolerant replication-based, and crash-tolerant erasure-coded). 5. Evaluation -- 2.5 pages. We're running on a cluster of 35 identical machines attached to a single gigabit switch. We're running four experiments: single client write throughput, read throughput, write latency, and read latency. We use a trivial synthetic workload. We'll generate four graphs and two tables of results. 6. Summary