Performance graphs hour/day/week

[stevehughes]

I have noticed that the performance graphs show inconsistent results depending on whether I select hourly, daily or weekly.

For instance - viewing the datastore latency for a particular VM for the past hour shows an average of around 1-2msec, with a spike to 40msec at 2:56 pm. If I change to 'past day' the spike drops to 5msec. This throws into doubt the usefulness of the 'past day' graph because I really don't know what it's showing me.

What's the go here?

Thanks,
Steve

[Shestakov]

Hello Steve,
That's an expected behavior. For past hour graph the values are shown with the best accuracy, i.e. we show performance values for every 20 seconds.
If we switch to past day/week, we cannot show all 20-seconds points, so they are combined to 5-minutes/1-hour points.
Thanks

[stevehughes]

Thanks for that info.

Is this a limitation just with how many point you can fit on the display, or is it a limitation in the way you store aging points in the database?

I would like to view the performance graphs at fine resolution for earlier times than just the past hour. For example this morning I noted a number of storage latency alerts from during the night, but the ''past day' view doesn't show them because of the averaging involved. I'd like to scroll back in time through the night, displaying in one-hour periods if necessary to pinpoint what was going on at the time. Any way to do this?

[Shestakov]

Is this a limitation just with how many point you can fit on the display, or is it a limitation in the way you store aging points in the database?

Both. You don't want to have 30k 20-sec performance points on a weekly graph and Veeam ONE does aggregation of historical data. It collects 20-second samples from ESXi and shows it during the day or so(depends on the settings), then it aggregates 20-second values to 5-minutes intervals and shows the average value and so on up to 2-hour averages. At the same time Veeam ONE remembers the max. sample values.

I would like to view the performance graphs at fine resolution for earlier times than just the past hour. For example this morning I noted a number of storage latency alerts from during the night, but the ''past day' view doesn't show them because of the averaging involved. I'd like to scroll back in time through the night, displaying in one-hour periods if necessary to pinpoint what was going on at the time. Any way to do this?

You can set custom interval to get the desired data on the graph.

[stevehughes]

I found a previous thread that describes what I'm looking for. https://forums.veeam.com/post232359.htm ... se#p232359

We have a similar sized infrastructure to the one described.

How do I make a change to the settings? Do I need to contact support?

[stevehughes]

Sorry I messed up that link. This is what I meant. You made reference to a global setting that can be changed at the expense of database size.

https://forums.veeam.com/post232359.htm ... se#p232359

[Shestakov]

Yes, it's recommended to change regkeys with the support team assistance.
Thanks!

[Talend45]

Hi,

I have Vmware vShpere 6.7 and I would like to monitor my virtual machines. For example, I would like to know the maximum usage over a period of the VM RAM and CPU.

For this I go to the tab “Monitor” then “Advanced” “Chart Options”. I choose “CPU” on the left menu and the Usage” counter.

When I choose «Last year» so from 31/01-2019 - 30/01/2020, I get in my results a maximum of 15 % at 29/01/2020.
However, when I choose «Last month» so from 31/12/2019 - 30/01/2020, I get a maximum of 25 at 29/01/2020.

I don’t understand these figures, why the value are not the same.

Can you explain me ?

[Shestakov]

Hi John and welcome to the community forums.
Please review the posts above. Feel free to ask additional questions.

[Talend45]

Hi Shestakov,

Thanks for your response.

But I don’t understand very well, I’m French

I want to know if my virtual machines is correctly defined (for example if I can reduce the memory or I must increase), as same for processor.

So for that, I must know from 2019/11/01 to now , the max % used of memory and CPU for each virtual machine.

The maximum is not the same that last month so I think it's not correct.

There is a solution ?

[Shestakov]

The thing is maximum for today or yesterday is taken from 20sec interval. But if you check 2019/11/01 performance counters are aggregated to 5-min intervals.
For your case I would use optimization reports to check if VMs are undersized or oversized.