This is the code for Doppel, an in-memory key/value transactional database. WARNING: This is research code, and does not include durability or RPC. Use at your own risk.
Doppel's design is described in "Phase Reconciliation for Contended In-Memory Transactions", presented at OSDI 2014.
Get the code:
go get github.com/narula/dlog
go get github.com/narula/prof
go get github.com/narula/wfmutex
go get github.com/narula/ddtxn
To run the tests, use go test.
Clone the list-cpus repo and add it to your path:
git clone git@github.com:narula/list-cpus.git
Add $GOPATH/bin and the list-cpus repo to your PATH environment variable
Run a benchmark:
cd $GOPATH/src/github.com/narula/ddtxn/benchmarks
go install ./single
python bm.py --exp=single --rlock --ncores=N