Compare two leveldb instances.
The levelDB iterator reads keys out of the database but never loads the entire database into memory. So essentially we have two streams that can be read in two threads concurrently.
As the streams are read, each value is hashed and the key is added to an unordered map with the hash as the value. Each thread will become mutually exclusive when it needs to write to the value map.
Both threads are a stream of ordered keys. Each thread will repeat the following steps in parallel...
Check if the key has been added A. The key has not been added, hash the value and add it to the value map. B. The key has been added, remove it from the map.
We want each database's iterator to produce duplicate keys and values so that
they can be deduplicated. After the streams have ended we can determine
if the deduplication process has been sucessful if the length of the map is 0.
Specify two database paths.
./ldbcmp ./fixtures/same_a ./fixtures/same_b