Here comes my fourth weekly report for the indexation and data persistance.
For the moment i'm still on the data persistance, and still on the same problem as the previous week.
I've tried a first solution which consist into a node splitting.
You just split a node into two node.
Then all node will detect this split and reput there data on this new node.
However, if the data are all the same then we go in a long recursion.
( Insert data, Too Much data, Split, warn for splitting, re-put data .. again to much data etc until a complet split)
Thats mean that if they all got the same prefix, then we are going to develop the all tree.
This week i try a new solution which is :
- One node detect the split and put data on all node underneath until the real node where data going to split.
- All other node detect that change and go at the end of the tree
For the moment there is no implementation of that since ive been stuck on another problem due to bug into OpenDht for the first part of the week
Anyway, see you next week.