Every performance model has its own components, but generally all models re composed of the following components.
These are interactive computer terminals at which a user sits and submits his transactions to the computer system. Here the user first submits a transaction and waits until he gets the response. After this point, he takes some time to enter the next transaction. This time interval is called "Think Time " and is denoted by "Z". It is usually measured in seconds. The performance model (closed) of a computer system also takes this think time into consideration. The number of terminals is denoted by "M".
This is the wait queue for any particular device. A transaction, upon its arrival at the CPU or a Disk for its execution may find the device busy. There may already be an executing transaction at the CPU or Disk. Therefore, this transaction queues up into the device wait queue and keeps awaiting its turn for execution.
This is the most important factor that affects the performance of a computer system. CPU Speed has a direct effect on the overall performance.
Commercially, the speed of the CPU is expressed by its cycle time (by len gth ..e.g.,20 nsec, or by rate ..e.g.,50 MHz.) It is also popularly expressed i n MIPS (millions of instructions per second), but since MIPS is totally depend ent on the instruction set of the computer, it may not be considered a good stan dard to compare performance of processors. Owing to this drawback, several standa rd programs (benchmarks) have been proposed for the purpose of performan ce comparision. The most popular of them are Whetstone, Dhrystone, Linpack, Perfect Club, SPEC, TPC benchmarks, etc.
The SPEC benchmark consists of CPU intensive programs in C and Fortran, and the results are given as SPECmarks.
Disk subsystems have a great affect on the performance of the computer systems, especially if the transactions are I/O intensive rather than CPU bound. The performance of these systems is measured in terms of certain times, characteristic of that particular disk, like latency time, seek time.
It is the time that elapses between issuing a request for data, and actually starting the data transfer. In a hard disk, this translates into the time it takes to position the disk's read/write head and rotate the disk so that the req uired sector or cluster is under the head. Latency is just one of many factors that influence disk access speeds.
It is the length of time required to move a disk drive's read/write head to a particular location on the disk. The major part of a hard disk's access time is actually seek time.
It is the speed at which a disk drive can transfer information from the drive to the processor, usually measured in megabytes/second. A high data transfer rate speeds up the process and hence improves the system performance.