Disk Aggregate and Headroom Performance

[SirLoinOfBeef]

Is there a way to calculate how many disks in total can be added to an aggregate and what the headroom performance is?

I have a calculation disks/raids + nodes but I have no idea what that means.

[bretta · Accepted solution]

If you are a customer, then your NetApp or partner sales team can help you with that. You can research some of the limits by using Hardware Universe (HWU): https://hwu.netapp.com . It depends on your platform and your storage OS version. NetApp doesn't currently expose our performance and capacity sizing tool to customers.

[SirLoinOfBeef]

Thank you. Is there a way to manually calculate the headroom?

[bretta]

Not that I'm aware of. Your partner or NetApp sales team could tell you the maximum throughput of the platform (with infinite disks) and a "generic" workload. Or they could use our sizing tool to get the exact number for your configuration and workloads.

[SirLoinOfBeef]

Does anyone else know? My senior engineer insists that I can use the data from ActiveIQ to create a graph but I just cannot find the data.

[bretta]

Can you be very specific about what you want a graph of? For example, what is the X and Y axis specifically? I'm the right person to help you, but I can't understand exactly what you would graph.

[SirLoinOfBeef]

Well, I know one is time intervals but I do not know what the other is.
I can see the Performance Capacity graph in ActiveIQ but my question is - is there a number that corresponds to "if I have 50% of Performance capacity used what is the 50% free - CPU? IOPS? Utilization? A combination of all three?"

[bretta]

Active IQ doesn't graph performance capacity. That's a system-level thing that blends CPU, disk, cache, and a few other things. We graph discreet data points (with one exception - CPU Utilization at the node level which I'll address later). Active IQ graphs:

Cluster level: Average IOPS, Network throughput

Node Level: CPU utilization, Latency, IOPS, Protocol IOPS, Network throughput

Aggregate level: Average throughput, Average utilization

Volume level: Volume latency, Volume IOPS

Note: None of these individually predict performance capacity or headroom because you may run out of disk before CPU or vice versa.

Note: There is one predictive graph which Active IQ shows, which is "peak performance" in the node-level CPU graph. This is predicted by ONTAP and is covered in the ONTAP documentation.

[SirLoinOfBeef]

Bummer.
How can I calculate headroom without using the cli commands?

[SirLoinOfBeef]

Are there any NetApp docs that show hows to calculate headroom?

[Ontapforrum]

Some kbs below incase you haven't seen them already. However, you will need to pull this data from CLI to some extent.

https://docs.netapp.com/us-en/ontap/performance-admin/identify-remaining-performance-capacity-task.html

https://kb.netapp.com/Advice_and_Troubleshooting/Data_Storage_Software/ONTAP_OS/What_is_%22PEAK_PERFORMANCE%22_line_in_Active_IQ%3F

https://kb.netapp.com/Advice_and_Troubleshooting/Data_Infrastructure_Management/OnCommand_Suite/What_is_Performance_Capacity_in_ONTAP_9_and_Unified_Ma...

https://kb.netapp.com/Advice_and_Troubleshooting/Data_Storage_Software/ONTAP_OS/Is_my_controller_overloaded%3F

[SirLoinOfBeef]

Thanks. I looked at those but was told not to use cli commands to calculate the headroom.