topic Re: High CPU with Data OnTAP EDGE in ONTAP Discussions

High CPU with Data OnTAP EDGE

TMADOCTHOMAS — Thu, 05 Jun 2025 05:34:17 GMT

Hello,

We are running the evaluation version of Data OnTAP EDGE on a server with 2 CPUs and 4 GB of RAM. Nonetheless, the CPU is spiking on the host. Has anyone else encountered this?

Re: High CPU with Data OnTAP EDGE

shatfield — Wed, 18 Jun 2014 04:57:46 GMT

Thats a little light, even for the eval. Edge fully consumes 2 cores and 4gb of ram, and the host needs some overhead. You really need 4 cores and 8gb ram in the host for Edge to run optimally.

Edge does not issue CPU HLT instructions on idle, so it is expected that the CPU cores used by the edge vCPUs will sit at 100%. In the non-eval version, 2 full cores of CPU cycles are reserved for Edge during install, along with a 4G RAM reservation (8GB for CDOT). DVadmin will also refuse to initialize the VM if power management is not disabled on the host. To get the best sense of how edge will perform it would be a good idea to replicate these conditions in the eval, and run it on a host with a supported raid controller and local disk configuration.

Re: High CPU with Data OnTAP EDGE

TMADOCTHOMAS — Wed, 18 Jun 2014 12:41:27 GMT

Thank you for the information and advice Sean!

Re: High CPU with Data OnTAP EDGE

TMADOCTHOMAS — Wed, 18 Jun 2014 13:13:25 GMT

Our server technician tried to add one core to each processor, and he received the following error:

ERROR: Incorrect CPU resources.

ERROR: Data ONTAP startup failed - initiating shutdown.

He then tried increasing memory from 4GB to 5GB and got the same error. When he removed the additional CPU and memory, the VM booted up fine. Is the evaluation locked to a certain amount of CPU and memory?

Re: High CPU with Data OnTAP EDGE

shatfield — Wed, 18 Jun 2014 15:13:00 GMT

The resource requirements are very specific. It must have 2 vCPU and 4GB of ram assigned to the VM. No more, no less. The production Edge models are the same. 7 mode Edge/Edge-T/Edge Premium are 2vCPU/4GB, The Clustered Ontap model requires 2vCPU/8GB.

Re: High CPU with Data OnTAP EDGE

TMADOCTHOMAS — Wed, 18 Jun 2014 15:26:05 GMT

Thanks Sean. I've reviewed the Data OnTAP Edge 8.2 documentation and noticed the following. From page 14 (emphasis mine):

-------------------------------------------------------------------

Server requirements

Your server should meet or exceed the following minimum requirements:

• Quad core, or two dual-core, 64-bit Intel x86 processors

• 2.27 GHz or greater processor speed

• 8 GB or greater memory

• 4 or more physical disks on the server

• Single Gigabit Ethernet network

• Hardware RAID - Must have a battery-backed write cache

-------------------------------------------------------------------

Then from page 15 (emphasis mine):

-------------------------------------------------------------------

Data ONTAP Edge resource requirements

You can install only one Data ONTAP Edge storage system per ESX host. The Data ONTAP Edge

virtual machine requires the following resources from the host server:

• 2 dedicated physical CPU cores

• 4 GB of dedicated memory

• 20 GB disk space for system disk images and virtual machine metadata

The Data ONTAP-v virtual machine reserves two CPU cores and 4 GB of RAM from the host. You

cannot change these settings or the storage system will not boot. Limiting resources is not supported

and has not been tested. It can negatively impact the I/O capability of the storage system and could

risk data loss.

-------------------------------------------------------------------

So I take it that the first list is minimum requirements for the server, but the EDGE VM itself must be configured as described in the second list. Is that accurate?

Re: High CPU with Data OnTAP EDGE

shatfield — Wed, 18 Jun 2014 15:55:34 GMT

Thats exactly right. The VM requirements are fixed, the server minimums include the extra overhead needed for the hypervisor and the dvadmin tool used in the non-eval version.

Re: High CPU with Data OnTAP EDGE

TMADOCTHOMAS — Wed, 18 Jun 2014 15:59:01 GMT

Excellent thanks for the help Sean!