You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: paper/paper.md
+3-3Lines changed: 3 additions & 3 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -169,17 +169,17 @@ While only two commits are shown in the result above, an actual export for this
169
169
The researcher could then perform an [analysis](https://github.com/vsoch/watchme-system/blob/master/data/watchme-task-analysis.ipynb) using the data collected.
170
170
As an example, here is a plot from such an analysis that tracks virtual memory usage of this author, recorded every hour, over two weekend days.
Interestingly, we can see a pattern that correlates with the activity of the author during the day. Virtual memory usage is low from the previous evening (1800 hours) through the early morning (0600 hours) and then rises sharply when the author starts to work.
175
175
It goes down briefly in the early afternoon when the author pauses for a break, and picks up afterward, stopping when it's time for dinner.
176
176
We see that the system's core temperature follows a similar trend:
177
177
178
-

178
+

179
179
180
180
We also see that the computer was briefly unplugged after the morning work session.
181
181
182
-

182
+

183
183
184
184
These kinds of metrics are interesting to answer research questions about system resources and behavior, and represent only the tip of the iceberg in terms of the scope of data that WatchMe could help collect.
185
185
For example, WatchMe would have interesting use cases for monitoring resources or jobs for HPC, or watching for changes in any kind of web resource (prices, climate data, API endpoints, etc.).
0 commit comments