Revision 33323f21117e5bb8cf72569302f109aabb387bb9 authored by Mark Callaghan on 14 September 2012, 19:35:02 UTC, committed by Mark Callaghan on 14 September 2012, 23:43:50 UTC
Summary: Reads via mmap on concurrent workloads are much slower than pread. For example on a 24-core server with storage that can do 100k IOPS or more I can get no more than 10k IOPS with mmap reads and 32+ threads. Test Plan: db_bench benchmarks Reviewers: dhruba, heyongqiang Reviewed By: heyongqiang Differential Revision: https://reviews.facebook.net/D5433
1 parent fa29f82
| File | Mode | Size |
|---|---|---|
| block.cc | -rw-r--r-- | 7.9 KB |
| block.h | -rw-r--r-- | 1.0 KB |
| block_builder.cc | -rw-r--r-- | 3.6 KB |
| block_builder.h | -rw-r--r-- | 1.7 KB |
| filter_block.cc | -rw-r--r-- | 3.3 KB |
| filter_block.h | -rw-r--r-- | 2.2 KB |
| filter_block_test.cc | -rw-r--r-- | 3.6 KB |
| format.cc | -rw-r--r-- | 5.2 KB |
| format.h | -rw-r--r-- | 3.1 KB |
| iterator.cc | -rw-r--r-- | 1.7 KB |
| iterator_wrapper.h | -rw-r--r-- | 2.0 KB |
| merger.cc | -rw-r--r-- | 4.8 KB |
| merger.h | -rw-r--r-- | 867 bytes |
| table.cc | -rw-r--r-- | 8.0 KB |
| table_builder.cc | -rw-r--r-- | 8.9 KB |
| table_test.cc | -rw-r--r-- | 24.3 KB |
| two_level_iterator.cc | -rw-r--r-- | 4.6 KB |
| two_level_iterator.h | -rw-r--r-- | 1.2 KB |
Computing file changes ...
