Time
14 hours 13 minutes
Difficulty
Intermediate
CEU/CPE
20

Video Description

Monitoring Resource Usage (continued) This lesson discusses the advanced performance charts. You can decide which chart type you want (line or stacked) which objects to look at (e.g. A VM, a host), counters, or statistics types. These methods show what has changed on a system at any given time and you can choose how often it updates. All of these are good methods of measuring system performance.

Video Transcription

00:04
Hello, Dean Pompilio again. We're now in part two of Lesson four in module 10.
00:09
We're still talking about monitoring our resource usage. So I mentioned advanced performance charts in the last are the first part of this lesson, and you'll get to see these in action when we do the lab. 17. After this is all, complete it. So when you look at the advanced performance charts, you can pick different chart types. Decide which objects
00:28
objects you want to actually look at,
00:31
and that could be, ah, VM. It could be a cluster. It could be a host.
00:35
Then we pick the counters. And as you'll see when we look at the interface, there's quite a huge list of available counters. Each of those categories can be expanded, and there's other items within a given category that you can look at.
00:50
Then we have. We'll talk about what the roll up function means and what the statistics type actually means. So the central statistics one thing I wanted to point out right away. These are not stored in the View central database. They're actually stored in a flat file on the host.
01:07
They are stored in the V center server's memory,
01:10
but that obviously is volatile. But the flag file in the host will hold the statistics so you can get your historical data over time.
01:18
So we have different time intervals to think about.
01:21
The default
01:23
that shows up,
01:23
uh, is should be past day,
01:27
but you can more get more effective information by changing that to real time. That basically shows you within the last hour.
01:34
What kind of
01:37
data are as has changed. So the frequency of updates in the past, our motors 20 seconds. So we get 100 80 of those in a one hour time period.
01:49
Uh, if you set the timer novel to past day now it updates. Every five minutes we get 288 of those.
01:56
And then in a day
01:57
past week, it's 30 minutes, 336 samples
02:00
past month. This two hour interval and then past year is one day terrible. Obviously, we get 365 samples within a year,
02:08
so this gives you a nice way to look at trends
02:12
within a real time scenario within last week, day, month, year, and this is a great way to see how your performance is doing over larger, longer periods of time.
02:25
When you look at the charts, we have two different options. Here you can look at a line chart
02:30
where everything that you're displaying shows a separately colored line.
02:34
I personally prefer that method over the stacked graph. Stack graph is useful, especially when you want to compare multiple V EMS at one time. Or if you want to see the CPU usage of Multiple PM's. It is good for that.
02:46
But it's a little bit difficult to read. You have to hover the mouse
02:50
over the area where you want to see what the value is, and then a little pop of window will show you what the value is. When you look at the line graph, it's clear what the values are because they're all superimposed over each other and you could just look at the structure of the chart. And it's pretty obvious
03:07
of the objects encounters
03:08
when you're looking at your advanced charts.
03:12
An object is this instance
03:14
or a aggregation of devices,
03:16
so this could be, ah, things like your overall CPU usage or overall memory usage, or drilling down a little further
03:24
into ah, more specific statistic
03:29
and then the counters are those statistics that you want to collect. So when we're we're setting up your advanced chart, you decide which counters you want to display.
03:38
There's a button to select all counters. There's a button to de select all counters,
03:42
and then you just pick and choose the ones you actually want to look at.
03:46
S o the stacks that I would want to collect for a CPU, for instance. I might look at the used time, the ready time ready. Time tells me how much time my
03:55
CPI was waiting before it's given some resource is to take some action to execute some instructions. Maybe I want to see the CPU percentage usage
04:05
for for memory. I might want to look at how much memories being swapped or the balloon driver activity
04:12
for my nick. Maybe the amount of members walk again or the number of package transmitted the number of packets received.
04:18
You can get statistics on packets that air dropped and so on.
04:24
Then we have to think about our statistics type.
04:27
So in this case,
04:30
I'm looking at the unit of measurement used during the
04:33
the measurement interval.
04:35
So I've got my different statistics types of If I'm concerned about the rate, First of all, this is the value over the current interval. So if I'm
04:45
if my current interval is within the past day, that means
04:47
that that the data frequency will be five minutes
04:51
and I might want to look at CPU usage within that five minute period.
04:57
So that will be one data point on the chart.
05:00
The Delta would be the change from the previous interval. So it's the difference between those two things again, I could look at something like CP ready time to see if my processors keeping up with the current demand or if it's suffering and performing worse than it was five minutes ago.
05:15
And then the absolute statistic type is the real absolute value independent of the interval. It doesn't really care whether you're looking at the past month of the past five minutes. It's gonna give you the absolute value that detects at that moment, for instance, how much memory is currently active. That might be something
05:35
you want to look at and this absolute
05:39
statistics type.
05:41
Then we have to think about the roll up option when you're when you're creating your chart, you'll see roll up at the bottom,
05:46
and this is a way to convert
05:49
the intervals that you're sampling into some other value. So
05:56
if I'm looking at the five minute
05:58
variable
06:00
so five minutes of the past, our stats get converted into one past day value.
06:06
It seems kind of confusing to think about it that way. But if I'm looking at something every five minutes for an hour,
06:14
that can, that can give me one value that gets that will be visible on the past day chart.
06:20
He's playing that one. Maybe it makes more sense.
06:23
I can look at 30 minutes of past day stats and that gets converted into one past week value.
06:30
So if you're looking at past day stats all day long and you see your 30 minute intervals,
06:35
UM, 30 minutes of those will give you one interval on your past week value so you can see how those stats roll up to create a new value for the larger time span intervals.
06:48
And then you also have to think about the logging level that you're using.
06:53
We can have no logging,
06:55
which would be really no good reason to do that. Other than the fact that you don't have space for storing logs, perhaps
07:00
then you could log on Lee errors,
07:03
which will give you a minimal level of logging. Maybe you're mostly concerned with errors anyway, so that's a good place to start. We can also logged warnings,
07:13
which will take more logs space. But these were things that you might want to fix before they become errors, right? So warnings are something that's not a showstopper, but it could develop into a more serious problem if you don't address it
07:26
and then we have info are informational logging.
07:30
This gives you a lot more data
07:31
for the moment to moment operations of what your host is doing what your VMS air doing. So this generates a lot more logging than we have verbose logging. This is sort of like a debug mode
07:45
where you're getting a tremendous amount of information about what the systems are doing, what your VMS air doing and then the largest logging level is trivia.
07:54
This gives you every possible laudable event,
07:57
and you might need this in certain circumstances. If you're doing development or troubleshooting, you might turn out a higher level of logging for a period of time. The trick is to remember that your that your are running an enhanced logging so that you can go back to just doing like errors and warnings, for instance,
08:13
so don't leave it in in in forward for both modes. Otherwise, you'll fill up your file system pretty quickly with megabytes and megabytes, gigabytes, even of logs.
08:24
Okay, so to recap, we talked a little bit more about what the advanced charts parameters are that you can adjust.
08:31
We looked at some of the different statistics for the different time intervals that are available.
08:35
We know where the log files are stored
08:39
are the statistics files are stored. Rather, we have are two different chart types.
08:43
We know what objects encounters are, and this is pretty obvious when you look at the interface and you'll see it in the lab.
08:50
And then we have statistic types based on whether or not we want to look at the current value, the absolute value or the value over the time interval. That's that selected. When we pick the statistics value,
09:03
then the roll ups. We can see how ah, certain number of short interval
09:07
data points. You rolled up into one data point for a longer interval.
09:11
And then lastly, we talked aboutthe log levels and why those are important for doing troubleshooting. But you want to keep them in a minimum and most other circumstances.
09:22
Okay, so lab 17 is the next task.
09:24
This lab will be creating some CPU activity using the CPU busier utility.
09:31
And then you'll get a chance to use the Web client to look at CPU utilization.
09:35
And then I'll reverse some of those changes to get ready for some subsequent labs.
09:41
All right, that concludes a lesson. Thank you.

Up Next

Virtualization Management

Our self-paced online Virtualization Management training class focuses on installing, configuring and managing virtualization software. You?ll learn how to work your way around the cloud and how to build the infrastructure for it.

Instructed By

Instructor Profile Image
Dean Pompilio
CEO of SteppingStone Solutions
Instructor