CCBench : Redesign and Implement Many Concurrency Control

This platform is undergoing rewrite in the repository below.
https://github.com/thawk105/ccbench_v2
Analysis paper using CCBench is below.
http://www.vldb.org/pvldb/vol13/p3531-tanabe.pdf

Installing a binary distribution package

On Debian/Ubuntu Linux, execute below statement or bootstrap_apt.sh.

$ git clone --recurse-submodules this_repository
$ cd ccbench
$ sudo apt update -y && sudo apt-get install -y $(cat build_tools/ubuntu.deps)

Prepare using

$ cd ccbench
$ "run some build_tools/(bootstrap*.sh) files"

• Processing of bootstrap.sh :
  Build third_party/masstree.
• Processing of bootstrap_mimalloc.sh :
  Build third_party/mimalloc.
  
• Processing of bootstrap_tbb.sh :
  Build third_party/tbb
  

Export LD_LIBRARY_PATH to appropriate paths.
Each protocols has own Makefile(or CMakeLists.txt), so you should build each.
Prepare gflags for command line options.

$ cd third_party/gflags
$ mkdir build
$ cd build
$ ccmake ..
  - Press 'c' to configure the build system and 'e' to ignore warnings.
  - Set CMAKE_INSTALL_PREFIX and other CMake variables and options.
  - Continue pressing 'c' until the option 'g' is available.
  - Then press 'g' to generate the configuration files for GNU Make.
$ make -j && make install

Prepare glog for command line options.

./autogen.sh && ./configure && make -j && make install

---

Data Structure

Masstree

This is a submodule.
usage:
git submodule init
git submodule update
tanabe's wrapper is include/masstree_wrapper.hpp

---

Experimental data

https://github.com/thawk105/ccdata

---

Runtime arguments

This system uses third_party/gflags and third_party/glog.
So you can use without runtime arguments, then it executes with default args.
You can also use runtime arguments like below.
Note that args you don't set is used default args.

$ ./cicada.exe -tuple_num=1000000 -thread_num=224

---

Details for improving performance

• It uses xoroshiro128plus which is high performance random generator.
• It is friendly to Linux vertual memory system.
• It uses high performance memory allocator mimalloc/tbd appropriately.
• It reduces memory management cost by our original technique.
• It refrain from creating temporary objects to improve performance as much as possible.
• It fixed bug of original cicada.
• It modifies almost protocols appropriately to improve performance.

---

Welcome

Welcome pull request about

• Improvement of performance in any workloads.
• Bug fix.
• Improvement about comments (doxygen style is recommended).
• Improvement of versatile.
• Extending CCBench
  • Reference materials : thawk105#7
• Extend tests.

---

Acknowledgments

Takayuki.T dedicates special thanks to ...

• Cybozu Labs Youth 8th term supported this activity. (2018/4/10 - 2019/4/10)
  
• Takashi Hoshino who is very kind advisor from Cybozu Labs Youth.
• Hideyuki Kawashima/Osamu Tatebe who is very kind supervisor.