We used [SAGIV] as our starting point for concurrent b-tree algorithms. However, we made significant changes and (we feel) improvements. We have particularly tried to simplify the algorithms to reduce the implementation and especially debugging effort. There are also a lot of complex details involved in building a complete implementation. The goal of this document is to describe and explain the major design decisions and any subtleties that arose.
|• B-tree Structure and Access:|
|• Buffer I/O and Free-List Management:|
|• Error Handling:|
|• Longer Value Fields:|
|• Unlimited Length Keys and Values:|
|• To Be Done:|