ArcGIS and Nutanix?

[DEWright_CA]

Has anyone else started looking or is working with the NUTANIX platform for VMWare and powering ArcGIS Services? I am intrigued by the mix of SSD to store the live access data while the HDD array maintains the less accessed data.

Are there any success stories out there?

Thanks;

David

[SeanMcLaughlin3]

Hi David,

I have interesting updates for you per your ask. Please contact me on sean.mclaughiln@nutanix.com or 703-980-3291

[DEWright_CA]

Hi Sean, I tried to email you and the message was bounced as unknown/undeliverable.

[Kelly_Estes]

Hi David, please try seanm@nutanix.com that one should work. You can also email kelly.estes@nutanix.com 

[RyanDanzey]

Esri Performance Engineering is working with Nutanix and will be publishing a blog and content in the near future.

[DEWright_CA]

Very happy to hear this; it has been years since I broached this topic at #ESRIUC and got blank stares...

[DanielThomas1]

Ryan,

Has there been any movement with Nutanix and ArcGIS Enterprise?  I am interested in deployment of datastores for enterprise into the Nutanix environment.

[DEWright_CA]

Hi Daniel, from my view we never saw enough traction from ESRI on this. But I did just complete my second deployment of a full NUTANIX environment for my Server environment and am happy to talk about it. We did not go with the Desktop Virtualization route, we are purely on the converged Server/Storage path.

Running VMWare/VSphere and Windows 2016/2019 VMs.

[DanielThomas1]

David,

Thank you for your response.  Did you deploy the datastores using AFS as 'OS' less file shares? Nutanix on the surface seems like a good fit for a highly available disk area and would be maintenance free if you go with Acropolis instead of using a Windows OS.

[DEWright_CA]

Sadly no, because of Auditing and Access-Control requirements we have to go with OS hosted shares for most items. Proper mapping of machines and distribution across your nodes helps a lot with reducing risk of contention and the needs to potentially have machines switch nodes. I went from a 3-node cluster to a 5-node, with one dedicated for cluster-class work.

The mixture of SSD and HDD has been a great performance boost; and maxing out your nodes with the most RAM you can is a huge boost for DB machines.