Last post you learned how to check the
performance data of your hosts, but that was an interactive way, there’s not
way to create a historical analyses based on that or even create graphical
representation with the data.
So, how do you export performance data
from ESXi for a more in depth analysis ?!?
First, let’s generate a .CSV file with
the performance data you want.
Make sure you configure esxtop to
gather just the metrics will want otherwise your .CSV file will be huge. (Check
the esxtop basics to learn how).
Then run esxtop in batch mode.
Example:
esxtop –b –d 10 –n 6 > /tmp/log.CSV
-b stands for batch mode
-d stands for delay, it means every X
seconds it will capture the data
-n is the number of interactions, it
means how many captures it will make
on the example above it will capture
the data 6 times, one at every 10 seconds, summarizing 60 seconds of data and
saving it on /tmp/log.CSV
Now that you have your file, copy it to
a desktop where you can read this data.
Once the file is in your desktop you
have a few options to read it:
-
Excel
-
Perfmon
-
Esxplot
Since it’s a .CSV file you could import
into Excel and work yourself with the data to create the graphics you want. It
can be challenge and also Excel has a limitation, the number of lines, so if
your file is big enough Excel could not handle that.
-
Open Excel
-
Select File/Import
-
Select CSV File and
then browse to the file with your data.
Another option is to use perfmon, most
Windows administrators are familiarized with this tool and would have no
problems with that. Also perfmon has
limitations, it could take a long time loading all the information from your
file.
-
Open Perfmon
-
Select Source tab
-
Change to log files
-
Cick Add then browse
to the file with your data.
Another great tool it esxplot, since it
has been build specific to work with ESX logs, it has the better performance
and can read enormous amount of data, once it’s installed you just need to
select your .CSV file to start seeing the data.
I bet you are anxious to start grabbing
some data for analysis, right ; )