The memory for write operations is now aligned to 4 KiB, which allows to later implement disks writes in DirectIO. It also benefits SMR writes by removing the need of memcpy to align the buffer with the underlying drive block size.
The previous approach was also resizing constantly the vector PacketStruct::packet, which implies frequent allocations. The new PacketStruct::alignedBuffer is now reused most of the time without constant allocations.
The memory for write operations is now aligned to 4 KiB, which allows to later implement disks writes in DirectIO. It also benefits SMR writes by removing the need of memcpy to align the buffer with the underlying drive block size.
The previous approach was also resizing constantly the vector PacketStruct::packet, which implies frequent allocations. The new PacketStruct::alignedBuffer is now reused most of the time without constant allocations.