which technology is being used for csv lookup replication (SH/indexer replication)

[pargupta1234]

I want to understand which technology is being used while CSV lookup bundle replication.
Is this whole CSV gets replicated or only changes in CSV get replicated?
I checked replication logs and found that not complete CSV gets replicated so why CSV lookup replication causes SH replication issue if the size of CSV exceeds 2 GB?
What is the problem with CSV lookup that it is not good for large lookup and everyone advice to move to KV store or DB connect?

[codebuilder]

The default limit for knowledge bundle replication is 2GB, it's not specific to lookups.
KV Store is generally recommended for lookups greater than 100MB because it is DB based (Mongo) and therefore much faster and more efficient.