In early linux systems, inodes were kept together on one part of a disk, while the corresponding data was scattered elsewhere on the disk. Later editions divided disks drives into group of adjacent cylinders called cylinder groups, in which each cylinder group contains inodes and their corresponding data. How does this new organization improve performance?
here's what i put down: The file manager must first read a sector to make sure that it has enough empty space in the sector to write the record to disk. To decrease the number of times such a read is likely to occur we can calculate how much space is in a sector the first time a read is called and only call read again after that sector can hold no more information. Continue subtracting from that first read until a record being inserted is too big to fit into a sector.