Data processing: measuring – calibrating – or testing – Measurement system – Performance or efficiency evaluation
Patent
1997-09-17
1999-09-21
Assouad, Patrick
Data processing: measuring, calibrating, or testing
Measurement system
Performance or efficiency evaluation
702186, 39520033, 39520054, 705 7, G06F 1900
Patent
active
059566620
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
The method relates to the operation of computer systems for measuring the performance in networked systems.
During the planning and installation of networked computer systems, it is expedient to evaluate or test the planned system in order to find out whether the planned system actually achieves the required performance. The number of instructions which can be executed per second by a central processing unit is for the most part inadequate for this purpose. Performance measurement with adjustable parameters is necessary for this.
To this end, use is made of a test installation and a plurality of load generators which generate a load which is similar to the planned application even though it is synthetic. It is thus possible to test whether the planned system can cope with a predetermined load situation.
It is, of course, the aim of a performance measurement to supplement this purely binary indication by obtaining information by means of quantitative indications, regarding the load profile, in particular the simultaneously active number of loads simulated by load generators, which utilizes a predetermined system to capacity. What becomes evident as a result, in particular, is the load which the system can cope with in addition to the specified load.
A first solution is to start up a number of load generators jointly and activate them successively until the loading limit is reached. However, it has turned out that the figures obtained by this means have resulted in incorrectly dimensioned systems. To date, this has been attributed to the fact that the load generators are just not realistic enough, and the performance figures obtained have been correspondingly corrected by estimation.
However, it has emerged that load generators are fully suitable for carrying out precise, realistic and comparable performance measurement if the method according to the invention is used to determine the performance figures.
SUMMARY OF THE INVENTION
The invention consists in first of all activating all of the load generators to produce an overload situation. After this, a plurality of load generators, preferably five, are in each case simultaneously deactivated until the overload situation disappears. After this, individual load generators are successively reactivated. As soon as the overload situation reappears the number of load generators before this point is the measure of the performance of the computer system with regard to the load profile defined by the load generators. The clients deactivated in the respective previous step are at the same time completely disconnected when further clients are deactivated.
BRIEF DESCRIPTION OF THE DRAWINGS
The features of the present invention which are believed to be novel, are set forth with particularity in the appended claims. The invention, together with further objects and advantages, may best be understood by reference to the following description taken in conjunction with the accompanying drawings, in the several Figures of which like reference numerals identify like elements, and in which:
FIG. 1 shows a block diagram of a measuring arrangement,
FIG. 2 shows a flow diagram of the sequence of events in the method.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
FIG. 1 shows a block diagram for performance measurement of client/server installations. A network 10, for example Ethernet or Token Ring, connects different computers. A computer 11 designated as the server is the computer to be surveyed, the intention being to determine its performance figure. For this purpose, further computers 12a, 12b and 12c, called clients, are connected to the network in such a way that they can communicate with the server. The number of clients is usually considerably larger than the number shown and can amount to two hundred or more in the case of very high-performance servers or in the case of a plurality of servers. A server can operate clients even in a plurality of different networks. The server is generally operated with a time-sharing operating system. The la
REFERENCES:
patent: 5440719 (1995-08-01), Hanes et al.
patent: 5539883 (1996-07-01), Allon et al.
patent: 5664106 (1997-09-01), Caccavale
patent: 5675731 (1997-10-01), Fuller
patent: 5691920 (1997-11-01), Levine et al.
patent: 5729726 (1998-03-01), Levine et al.
patent: 5742819 (1998-04-01), Caccavale
Drakopoulos, "Analysis of a Local Computer Network with Workstations and X Terminals", IEEE., Mar. 1992.
Philips, "Prophesy Brings Solid Network Simulation on a Tight Budget", PC Week, Oct. 1994.
USENIX Association, Proceedings of the 6.sup.TH Systems Administration Conference, (LISA VI), Oct. 1992, H.L. Stern et al, "NFS Performance and Network Loading", pp. 33-38.
Proceedings of the Summer USENIX Conference, Jun. 1993, M. Wittle et al, "LADDIS: The Next Generation in NFS File Server Benchmarking", pp. 111-128.
Proceedings of the IEEE, vol. 81, No. 8, Aug. 1993, P.M. Chen et al, "Storage Performance-Metrics and Benchmarks", pp. 1151-1165.
Elektrisches Nachrichtenwesen, vol. 59, No. 1/2, (1985), G. Morales Andres et al, System 12 uberlastabwehrstrategie, pp. 74-79.
Hemker Josef
Pfeifer Dirk
Assouad Patrick
Siemens Nixdorf Informationssystem Aktiengesellschaft
LandOfFree
Method for load measurement does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for load measurement, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for load measurement will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-91323