This International Standard defines how user oriented performance of computer-based software systems (CBSS)

may be measured and rated. A CBSS is a data processing system as it is seen by its users, e.g. by users at various

terminals, or as it is seen by operational users and business users at the data processing center.

A CBSS includes hardware and all its software (system software and application software) which is needed to

realize the data processing functions required by the users or what may influence to the CBSS's time behaviour.

This International Standard is applicable for tests of all time constrained systems or system parts. Also a network

may be part of a system or may be the main subject of a test. The method defined in this International Standard is

not limited to special cases like classic batch or terminal-host systems, e.g. also included are client server systems

or, with a broader comprehension of the definition of ?task', real time systems. But the practicability of tests may be

limited by the expenditure required to test large environments.

This International Standard specifies the key figures of user oriented performance terms and specifies a method of

measuring and rating these performance values. The specified performance values are those which describe the

execution speed of user orders (tasks), namely the triple of:

- execution time,

- throughput,

- timeliness.

The user orders, subsequently called tasks, may be of simple or complex internal structure. A task may be a job,

transaction, process or a more complex structure, but with a defined start and end depending on the needs of the

evaluator. When evaluating the performance it is possible to use this International Standard for measuring the time

behaviour with reference to business transaction completion times in addition to other individual response times.

The rating is done with respect to users requirements or by comparing two or more measured systems (types or

versions).

Intentionally no proposals for measuring internal values, such as:

- utilisation values,

- mean instruction rates,

- path lengths,

- cache hit rates,

- queuing times,

- service times,

are given, because the definition of internal values depends on the architecture of the hardware and the software of

the system under test. Contrary to this the user oriented performance values which are defined in this International

Standard are independent of architecture. The definition of internal performance values can be done independently

from the definition of user oriented performance values. They may be used and can be measured in addition to the

user oriented performance values. Also the definition of terms for the efficiency with which the user oriented values

are produced can be done freely. In addition this International Standard gives guidance on how to establish at a

data processing system a stable and reproducible state of operation. This reproducible state may be used to

measure other performance values such as the above mentioned internal values.

This International Standard focuses on:

- application software;

- system software;

- turn-key systems (i.e. systems consisting of an application software, the system software and the

hardware for which it was designed);

- general data processing systems.

This International Standard specifies the requirements for an emulation (by a technical system - the so-called

remote terminal emulator (RTE) - of user interactions with a data processing system. It is the guideline for precisely

measuring and rating the user oriented performance values. It provides the guideline for estimating these values

with the required accuracy and repeatability of CBSSs with deterministic as well as random behaviour of users. It is

also a guidance for implementing a RTE or proving whether it works according to this International Standard.

This International Standard provides the guideline to measure and rate the performance of CBSS with random user

behaviour when the accuracy and repeatability is required. It specifies in detail how to prepare and carry out the

measurement procedure. Along with a description of the analysis of the measured values, the formulas for

computing the performance value and the rating value, are provided.

This International Standard also gives guidance on:

- how to design a user oriented benchmark test using a:

* transaction oriented workload,

* batch oriented workload,

* or transaction and batch mixed workload.

It specifies:

- how to describe such a workload,

- how to perform the measurement procedure,

- how to rate the measured results.

This International Standard is of interest to:

- evaluators,

- developers,

- buyers (including users of a data processing system),

- system integrators

of CBSSs.

NOTE 1 The field of application of this International Standard may be extended to include the following aspects.

Workloads fulfilling the specifications of this standard and having a sufficiently general structure may be used as standard

workloads. They may be used to measure and rate performance of data processing systems used in specific fields. E.g. a

standard workload for word-processing may be used to compare the time efficiency of different software products or

different versions of the same product running on the same hardware system. Such a standard workload may also be

used if always applying the same application software version and the same hardware to compare the efficiency of the

system software. When applying the same application software and workload to different systems, consisting of hardware

and system software, as normally sold by system vendors, the efficiency of the data processing systems may be

compared with respect to the application and workload used.