LevelDB is fast in writing but has O(n) performance in reading, so the longer the uptime, the more disadvantageous the performance. Also the bloom filter has a limit in accuracy as more items are added.
So, implement Trie to support faster search.
I'm thinking of running Trie in memory by default, and storing Trie in levelDB if the storage option is turned on. A process to load Trie from levelDB on restart should also be added.
AS IS)
LevelDB is fast in writing but has O(n) performance in reading, so the longer the uptime, the more disadvantageous the performance. Also the bloom filter has a limit in accuracy as more items are added.
So, implement Trie to support faster search. I'm thinking of running Trie in memory by default, and storing Trie in levelDB if the storage option is turned on. A process to load Trie from levelDB on restart should also be added.