This page contains links to older performance measurements (see the home page for the most recent and significant measurements).
- Metrics Files (generated during tests of crash recovery)
- Perf benchmarks (microbenchmarks measuring various low-level operations on a single machine, such as atomic increment)
- clusterperf benchmarks (benchmarks run on a cluster to measure basic things such as read and write latency and throughput)
- Preliminary multiRead benchmarks (June 2012) (Better multi-op measurements are included in clusterperf benchmarks after this)
- Latency Patterns in Infiniband (May 2012)
- RPC Latency Profile (January 2012)
- SSD Experiments (July 2011)
- Waking up worker threads (July 2011)
- Mellanox Performance Data (May 2011)
- RPC Measurements May 2011
- network failure latency (Feb 2011)
- disk bandwidth (Feb 2011, updated Aug 2011)
- memcpy (January 2011)
- Example of Electric Cloud Performance Metrics
- Recovery Performance
- Mario's talk at the EPIC meeting pptx (October 2010)
- Rdtsc and Synchronization (October 2010)
- Infiniband Verbs Performance (August 2010)
- ZooKeeper Performance (June 2010)
- RPC Performance Numbers (December 2009)
- Hash Table & Multi-Level Lookup Performance (November 2009)