-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Optimization of the proposed protocol #8
Labels
Comments
Potential race condition at Need to check again before executing. Use something like Or an alternative to detect conflicts can be:
|
Current Comparison with OCC (Non-persistence)
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Problem
Currently, the PNVM is almost 2x slower than the OCC version with moderate conflict (4 out of 8 transaction types are conflicting transactions)
Here is a footprint of the time spent at executing a transaction from profiling:
The bottlenecks are :
Most of the things above can be supported by lock-free concurrency control.
Solution
crossbeam will be the library to use to support lock-free concurrency control. (It leverages on epoch memory deallocation)
For example, each TxnInfo can be a HashMap of
<Pid, ArcCell<PieceState>>
, and the writer will not be blocked by readers, and it will update the PieceState immediately (atomically), and readers afterwards will get to the updated version, while the old copy will be GCed after the last reader exits.The text was updated successfully, but these errors were encountered: