Friday, 3 January 2020

Network Switch Diagrams for a Minimum NetApp HCI Deployment*

*The minimum NetApp HCI Deployment is:
- 4 x H410S Storage Node
- 2 x H410C Compute Node (using two-cable configuration)

Here I present some diagrams for a minimum NetApp HCI Deployment. It’s not a cable diagram, instead I label ports with numbered and coloured discs. Essentially this is a blog post of 4 pictures:

i) Legend
ii) 1 GbE Management Switches
iii) 10/25GbE Network Switches
iv) NetApp HCI Chassis (with H410S + H410C)

Image: i) Legend

Image: ii) 1 GbE Management Switches

Image: iii) 10/25GbE Network Switches

Image: iv) NetApp HCI Chassis (with H410S + H410C)
  
The key to success with running the NetApp Deployment Engine (NDE) is to fully understand the:

Specifically, for this deployment:
> Network and switch requirements
- Need minimum of 3 VLANs: Management, vMotion, Storage (iSCSI).
- Additional VLANs: VM Network(s) [Initial VMs can use the Management network]
Note: Easiest setup if:
i) Storage Nodes 1GbE Management Connectivity Ports are Access Ports in the Management VLAN.
ii) All 10/25GbE ports have Management VLAN as the native VLAN (all 10/25GbE ports are trunk ports).
- All switch ports connected to NetApp HCI nodes must be configured as spanning tree edge ports.
- You must configure jumbo frames on the 10/25GbE switch ports
> Networks ports used by NetApp HCI
- If you want the Management Node (mNode) to reach external services - for instance, for ActiveIQ - review the table.
> Network cable requirements
- 14 Cat 5e/6 cables with RJ45 connectors
- 12 Twinax cables with SFP28/SFP+ connectors {OR} 6 switch side SFPs for 10/25GbE, 6 NetApp HCI side SFPs for 10/25GbE
> IP address requirements
- 2 x Compute nodes: 2 x Management IPs, 4 x Storage IPs, 2 x vMotion IP
- 4 x Storage nodes: 4 x Management IPs, 4 x Storage IPs
- Storage Cluster: 1 x Management IP, 1 x Storage IP
- VMware vCenter: 1 x Management IP
- SolidFire mNode: 1 x Management IP, 1 x Storage IP
- Temporary IP to run NDE: 1 x Management IP (applied to first Storage Node)
- TOTAL: 10 x Management IPs, 10 x Storage IPs, 2 x vMotion IPs
> Configuring LACP for optimal storage performance
- You have configured the switch ports connected to the 10/25GbE interfaces of NetApp HCI storage nodes as LACP port channels.
- You have set the LACP timers on the switches handling storage traffic to “fast mode (1s)” for optimal failover detection time.
> DNS and timekeeping requirements
- If you are deploying NetApp HCI with a new VMware vSphere installation using a fully qualified domain name, you must create one Pointer (PTR) record and one Address (A) record for vCenter Server on any DNS servers in use before deployment.
- If you are deploying NetApp HCI with a new vSphere installation using only IP addresses, you do not need to create new DNS records for vCenter.
- NetApp HCI requires a valid NTP server for timekeeping. You can use a publicly available time server if you do not have one in your environment.

Note: The 2-cable method requires VMware distributed switching (VDS) and the VMware Enterprise Plus licensing.

If all the requirements are met, NDE runs faultlessly and your environment is deployed in 30 minutes or so!
Environment here being: 4 x Storage Nodes. 2 x Compute Nodes (VMware ESXi). 1 x vCenter. 1 x SolidFire mNode.

No comments:

Post a Comment