One VM per CSV?

[TheVMinator]

I'm using the metric graph widget.  I can't see a way to get data from more than one VM at a time on a single CSV.  For example, CPU Demand (MHz) for vm1 and CPU Demand (MHz) for vm2 on the same csv file.   I can only download CPU Demand (MHz) for csv for one VM at a time. Two separate csv files, then have to combine them.  Is this a design limitation?  Someone please Tell me there's a better way.....

[mark_j]

The vCOps product is intended to be the end consumer/repository for the data, not so much to be a performance data provider for another solution. It does everything you'd pretty much need from a monitoring/capacity planning perspective. So if you want to export data FROM the vCOps appliance TO an external solution (e.g. SCOM), there's isn't an OOTB way to accomplish this. You could use something like the HTTP post adapter to do this, as others have, however that is a development effort or PSO engagement to make it happen.

[TheVMinator]

Thanks for the input.  This does sound quite limiting.

My strategy here was, get output from multiple VMs on one spreadsheet for CPU Demand (MHz).  Afterward I can add the values, average the values, compare the differences, or whatever I want to.

So I can't do that with vCOps because I can only download the data from one VM per CSV.

However suppose I just want to

1.  get the Average CPU Demand (MHz) (95th percentile) between two VMs.

2. have vCOps calculate this

3.  Download the value into a csv

Can I do this?  For example, can I create a supermetric representing the average, put that supermetric on the "metric graph" widget, then download the supermetric result to csv?  The supermetric would then represent the average between the 2VMs, and if it appeared on one graph, I could still download it to csv and have what I need.

Is this possible?

[mark_j]

You can do what you want via the formulas / SMs / and dashboards within vCOps. You don't want to avg the values of the 2 VMs together, then get the 9th percentile. You want to get the 95th percentile of the VM individually or get the 95th percentile of the SUM of the CPU demand for both VMs. Getting a 95th percentile of an AVG is going to smooth your data too far and negate the benefits of the 95th percentile analysis.

So.. use the data distro to display the 95th perc of cpu demand for a CPU or SUM of CPU for VMs, observe it, and there you go. There is no CSV value because 95th percentile is a single number.. not a time series. If you wanted the raw data, just go to a normal metric graph and export the raw data points and manipulate the data yourself.

If you want to do an export of the metric data on a large scale, see using the repository adapter as discussed in the custom ui admin guide. However, this is likely going to be overkill for the few resources/datapoints your looking for.

[TheVMinator]

mark.j wrote:

You can do what you want via the formulas / SMs / and dashboards within vCOps. You don't want to avg the values of the 2 VMs together, then get the 9th percentile. You want to get the 95th percentile of the VM individually or get the 95th percentile of the SUM of the CPU demand for both VMs. Getting a 95th percentile of an AVG is going to smooth your data too far and negate the benefits of the 95th percentile analysis.

So.. use the data distro to display the 95th perc of cpu demand for a CPU or SUM of CPU for VMs, observe it, and there you go. There is no CSV value because 95th percentile is a single number.. not a time series. If you wanted the raw data, just go to a normal metric graph and export the raw data points and manipulate the data yourself.

If you want to do an export of the metric data on a large scale, see using the repository adapter as discussed in the custom ui admin guide. However, this is likely going to be overkill for the few resources/datapoints your looking for.

OK thanks.  So I'm happy getting the raw datapoints from the metric graph widget.  I can do that now already and get a list of all the data points in a single csv for a single VM.  Excel can average them no problem.  The problem is how to get those datapoints for multiple VMs on the same csv.  I'm still not completely clear as to how I can solve this problem. Even if I create a supermetric and attach the supermetric to a resource tag containing multiple VMs, when I go back into the metric graph widget, I still can only make the supermetric give me data points on a single VM at a time.  I cannot seem to make it give me data points which represent the SUM of CPU demand for more than one VM so that I can download it in one CSV.  Are you saying I should be able to make the SUM of CPU demand more than one VM appear on the metric graph so I can download that SUM into a csv?

[mark_j]

Q: Are you saying I should be able to make the SUM of CPU demand more than one VM appear on the metric graph so I can download that SUM into a csv?

A: Sure, you could do this. And you'd do this using a supermetric to sumN(  <rk#:a#> ,1) the VMs. Obviously you'd need to throw these VMs in a group and apply the SM to this group to accomplish this.

[TheVMinator]

OK thanks.  When you say "throw them into a group", that is perhaps where I'm stuck.

Do you mean going into enviornment overview -> Manage resource tags - > Add tag

and adding a resource tag,

Then adding the VMs to a resource tag?

Where I am I creating this "group"?

[mark_j]

Create a group in the Advanced UI. It will then show up in the Custom UI, where you'll assign a SM to it.

The Group type you create the group under in the Advanced UI will appear under that specific Group type ResourceKind in the Custom UI. Just treat it like a container in terms of apply the SM to it.

[TheVMinator · Accepted solution]

OK thanks.  So when I create this supermetric, I select my group that I created in the Admin UI.

Do I then select the "Total Demand (MHz) under "Metrics" or under "Attribute Kinds"?

Am I actually doing a sumN() on the "Total Demand (MHz)"?

Should I add this SM to an attribute package?

When I create this supermetric, I can assign it to the group I created in the supermetric editor.  However I still can't make it appear on the "Metric Graph" widget for the group that was created in the Admin UI.  I can only make it appear for one VM at a time.

[mark_j]

As a rule of thumb, use the existing attributes/metrics where you can before creating SMs. Sometimes the metric you're seeking is already there on container resources such as groups. If you're interested in specifically CPU Demand (Mhz), you're in luck. This metric is already aggregated for a Group ResourceKind.

This is the metric:     Virtual Machine > CPU > Total Demand(MHz)

You're losing me a little bit on your last two sentences, but it seems like you're having a problem applying the supermetric to a resource. As a reference, a supermetric by itself in the editor window is merely a definition of a formula. It needs to be added to a supermetric package, then that package needs to be assigned to a resources to be applied. The SM will then be calculated with the next collection.

[TheVMinator]

OK thanks