Check Ahead
Why you need it?
In the current File mode, transaction records are written to disk in an append-only manner. Upon restart, the system must scan and reload all persisted records to rebuild its state, resulting in lengthy recovery times and high disk I/O consumption.
Furthermore, since all data is stored in memory, the system's capacity is constrained by the amount of available RAM.
Integrating RocksDB allows us to separate hot and cold data more effectively. Frequently accessed data can remain memory-resident, while colder data is persisted on disk. This not only reduces recovery time and memory pressure, but also enables much better scalability without sacrificing throughput.
How it could be?
No response
Other related information
No response
Check Ahead
I have searched the issues of this repository and believe that this is not a duplicate.
I am willing to try to implement this feature myself.
Why you need it?
In the current File mode, transaction records are written to disk in an append-only manner. Upon restart, the system must scan and reload all persisted records to rebuild its state, resulting in lengthy recovery times and high disk I/O consumption.
Furthermore, since all data is stored in memory, the system's capacity is constrained by the amount of available RAM.
Integrating RocksDB allows us to separate hot and cold data more effectively. Frequently accessed data can remain memory-resident, while colder data is persisted on disk. This not only reduces recovery time and memory pressure, but also enables much better scalability without sacrificing throughput.
How it could be?
No response
Other related information
No response