Browse Prior Art Database

Improving Performance of Trace Analysis using Visual Editing Functions

IP.com Disclosure Number: IPCOM000115030D
Original Publication Date: 1995-Mar-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 4 page(s) / 337K

Publishing Venue

IBM

Related People

Lehr, TF: AUTHOR

Abstract

"Tracing" a computer system is the recording of the events or states which the system executed. Trace chronologies are used to analyze and debug the performance of computer systems. When tracing time-stamps the events, the chronology can ordered visually as a time-line marked with various icons, colors and shapes to represent the events contained in the trace record. There are a number of visual tools which attempt to order trace data in these ways as a means for giving performance analysts powerful ways to study the casual relationships of trace events.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 43% of the total text.

Improving Performance of Trace Analysis using Visual Editing Functions

      "Tracing" a computer system is the recording of the events or
states which the system executed.  Trace chronologies are used to
analyze and debug the performance of computer systems.  When tracing
time-stamps the events, the chronology can ordered visually as a
time-line marked with various icons, colors and shapes to represent
the events contained in the trace record.  There are a number of
visual tools which attempt to order trace data in these ways as a
means for giving performance analysts powerful ways to study the
casual relationships of trace events.

      These tools are especially powerful when the traces contain, as
they often do, the equivalent of hundreds or even ten-of-thousands of
pages of textual information.  In fact, without such tools, only
small fractions of such traces are ever studied with any success.
The ability to "see the forest instead of the trees" is the sine qua
non for turning to trace visualization tools when analyzing trace
data of very large traces (PieScope goes beyond  other tools by
allowing user to see the forest as well as the trees).

      PieScope is a visual performance analysis tool which builds
various time-lines of a trace.   Along the time-lines it draws
colored bars representing begin, end and duration of interesting
events in the trace.  Although, PieScope is quite flexible in how it
represents trace data, its basic view consists of several time-lines
representing different "threads of execution" contained in the trace.
One might compare a thread of execution to one of several runners in
a road race.  If one imagines that the New York Marathon is traced by
time-stamping the arrival of each runner at various points in the
race, one could draw a time-line per runner with marks (icons,
colors, etc) along the time-lines indicating those arrivals.  If
PieScope were to visualize these time-lines it would place them in
parallel to one one another for easy comparison of the runners'
performances.

      Imagine also that the trace recorded when the runners arrived
at and left refreshment stations, when they took refreshment or
passed under cooling showers.  Imagine the trace also noted when the
runner's competitive position changed (e.g., the runner moved from
third to second place).

      Every trace visualization tool makes decisions on how to
initially categorize trace events when the tool parses the trace
data.  Assume that, in the case of the New York Marathon, PieScope
initially creates five categories:  1) "took refreshment," 2) "took a
shower," 3) "in 1st-3rd place," 4) "in 4th-6th place," and 5) "7th
place or  below."

      PieScope could represent category 1 as, say, the color blue and
draw blue bars on each time-line during the periods when a runner had
taken refreshment.  If category 3 might be represented by the color
red.  PieScope would then draw red bars on each time,...