Sequential I/O is operation in which adjacent data is accessed on a disk drive. Track to track seek is performed and thus greater throughput is provided.
And in case of random I/O, disk head reads data scattered on various parts of disk. Different tracks are read that are not adjacent, so random movement of head causes degrade in performance.
Keeping in view the major characteristic of sequential I/O and random I/O, it is important to design database disk storage in such a way that maximum sequential I/O may be performed. Transactional log is a major example of sequential I/O in SQL Server environment. You should place a heavily used log file on one disk. And it is good approach to use just only one default log file. If you place multiple log files on same disk the access type will be changed from sequential to random I/O. Practically it is not possible to achieve all I/O as sequential I/O in a server based environment. Clients request different operations for different pieces of data. So the only solution is to map your data on multiple disks so that random access should be minimized.
Click here to read all posts related to SQL Server storage systems and I/O operations