You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I would like to ask why fixed size blocks are used. Considering that this WAL implementation is used as bitcask storage log, the waste of storage space due to padding + record type is significant. I noticed that RocksDB wal also has a similar structure. Compared to directly appending records, what are the advantages of doing it this way?
The text was updated successfully, but these errors were encountered:
When you read a large wal file from the begining to end, read a fix-sized block is more efficient. Normally the size is equals to the page cache size.
if one record is small, like 1KB, you will read 32 tims to load 32 records.
but if the block size is 32k, we can read them all once, which will reduce the syscall(fread).
I would like to ask why fixed size blocks are used. Considering that this WAL implementation is used as bitcask storage log, the waste of storage space due to padding + record type is significant. I noticed that RocksDB wal also has a similar structure. Compared to directly appending records, what are the advantages of doing it this way?
The text was updated successfully, but these errors were encountered: