Electrical computers and digital processing systems: multicomput – Computer network managing – Computer network monitoring
Reexamination Certificate
1997-11-26
2001-01-23
Maung, Zarni (Department: 2757)
Electrical computers and digital processing systems: multicomput
Computer network managing
Computer network monitoring
C709S202000, C709S203000, C709S223000
Reexamination Certificate
active
06178449
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
This invention generally relates to computer systems, and more specifically relates to an apparatus and method for measuring transaction time in a computer system.
2. Background Art
The development of the EDVAC computer system of 1948 is often cited as the beginning of the computer era. Since that time, computer systems have evolved into extremely sophisticated devices, and computer systems may be found in many different environments. Since the dawn of the computer age, the performance of computers has been measured to determine how well the computer performs certain tasks. One measure of computer performance is response time, which is used herein to broadly define the time it takes the computer to perform a specific task or transaction. Generally, the response time that is of most interest is the user-perceived response time. For a single user on a single stand-alone computer system, the user-perceived response time is virtually the same as the time it takes the computer to perform the desired task or transaction. However, as multiple users and computers are added, the user-perceived response time is different than the processing time for the task on a single computer.
In the early days of computers, one or more relatively powerful and expensive computers could be shared by many users. Referring to
FIG. 1
, a number of computer terminals
130
were typically connected to a single computer
110
known as a “host.” These computer terminals
130
are commonly known as non-programmable workstations (i.e., “dumb terminals”) because they simply display information transmitted to it by the host, and lack any processing power to perform local tasks. One example of a very well-known computer terminal is the IBM 5250, which displays alphanumeric text in row and column format. In a computing environment
100
with a host
110
and one or more terminals
130
(referred to herein as a host computing environment), a communications controller
120
was typically included to facilitate the communications between the single host
110
and multiple terminals
130
. In a host computing environment
100
, software applications run on host
110
, and display information is transmitted by host
110
via communications controller
120
to terminals
130
. In this manner a user sitting at a particular terminal
130
may start an application on host
110
, and host
110
will then display an appropriate screen to the user at the appropriate terminal
130
. The user may then enter data in response to the displayed screen, if required.
A host computer and its dumb terminals communicate using an architected data stream that determines the action to be taken when certain characters are transmitted back and forth. When the user desires to perform a task or transaction, the user inputs the appropriate information on the screen to start the task, and presses the enter key to send this information to the host in a data stream. The host takes the information, processes the information appropriately, returns data to the dumb terminal in a data stream (that is typically displayed on the screen), and then unlocks the keyboard for further input by the user.
Determining the user-perceived response time in a host computing environment is relatively straightforward by monitoring the data stream between the host and the terminals. The communications controller
120
monitors the data stream to determine when and how to pass the data between host
110
and terminals
130
. As a result, integrating a response time measurement mechanism
160
into communications controller
120
is relatively straightforward. Response time measurement mechanism
160
includes a response time collector
140
and a response time list
150
. A response time collector
140
determines the user-perceived response time, and stores this information in a response time list
150
. The response time collector
140
starts a timer when the user presses the “Enter” key on his or her terminal, and stops the timer when host
110
sends the “keyboard unlock” message back to that terminal. Response time collector
140
then stores the response time in response time list
150
. The contents of response time list
150
may be accessed by host
110
as needed. Response time monitoring mechanism
160
thus provides a simple way to accurately determine the user-perceived response time in a host computing environment
100
. However, determining the user-perceived response time in a network computer environment presents greater challenges.
The widespread proliferation of computers prompted the development of computer networks that allow computers to communicate with each other. With the introduction of the personal computer (PC), computing became accessible to large numbers of people. Networks for personal computers were developed that allow individual users to communicate with each other. In this manner, a large number of people within a company, for example, could communicate at the same time with a software application running on one computer system. A computing system that includes multiple computers communicating over a network is generically referred to herein as a network computing environment. One example of a suitable network computing environment
200
is shown in
FIG. 2
, and includes a server computer system
210
communicating over a network with multiple clients
230
.
Today, computer networks are ubiquitous, and great time and effort is being expended to increase the performance of computer networks. One aspect of performance is the user-perceived response time. However, with computers in a network computing environment, known network protocols do not provide any uniform mechanism for determining user-perceived response time. So important are user-perceived response times that some consultants are now basing their fees on the improvement in user-perceived response times they are able to achieve. Therefore, measurement of user-perceived response time is critically important. The problem of accurately measuring user-perceived response time in a network computing environment is complicated by the fact that multiple computers on the network may process sub-parts of a task or transaction in parallel, so the different computers that perform the various sub-tasks will not know when the entire task or transaction is complete. The focus of response time measurements in a network computing environment have been on measuring transaction times. Transaction time is the time required for a transaction to go from start to finish, regardless of how the various steps are accomplished. For the discussion herein, the term transaction time is understood to be the user-perceived response time.
One possible solution to monitoring user-perceived response time is illustrated in
FIG. 2
, and would place a response time measurement mechanism
260
in the server computer system
210
. Response time measurement mechanism
260
could have a response time collector
240
and a response time list
250
. Response time collector
240
would actively monitor and interrogate each client
230
to determine response times, and would store these response times in response time list
250
. This solution, however, takes considerable processing power and excessive overhead, which would significantly degrade system performance. For this reason, this solution has not been implemented in large-scale networks.
The prior art methods of measuring response time in a network computing environment have excessive overhead, and are not easily implemented. Therefore, there existed a need to provide a response time monitoring mechanism that efficiently collects response times in a network computing environment without introducing excessive overhead.
DISCLOSURE OF INVENTION
According to preferred embodiments of the present invention, a transaction time measurement mechanism has a transaction time manager running on a server computer system, a transaction time agent running on a client computer system that is coupled to the server computer system vi
Forman Ron Frank
Pechacek John M.
Schwane Walter H.
International Business Machines - Corporation
Le Quoc-Khanh
Martin Derek P.
Martin & Associates, L.L.C
Maung Zarni
LandOfFree
Apparatus and method for measuring transaction time in a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for measuring transaction time in a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for measuring transaction time in a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2515863