At first, I was going to walk through a VxRail install and talk about what was needed, but there’s already quite a few different resources highlighting that. What I wanted to do was build more on what to watch for with regards to an install. VxRail, today, is installed via Dell EMC’s or a certified Partner’s professional services organization. This post is probably most relevant to those parties, but I also think our customers will appreciate the content.
If you’ve ever seen a demo of a VxRail install, you’ve seen the following screens:
For the first screen, it’s just the beginning screen. If this isn’t seen, there could be a few things to look into. The first is to validate that IPv6 is enabled and IPv6 multicast traffic is allowed. The next thing we’d look at is whether or not the VxRail Manager VM was powered up on the nodes.
If you notice, the first node in my config has a little house icon next it. That icon signifies which node has won the election process and will be the node that will run the VxRail Manager VM to facilitate the install of the appliance. At this point, all the nodes are booted into ESXi off their internal ESXi boot device – the device will vary based off the hardware platform. If on the screens showing the nodes the system discovered isn’t populating, my best guess is that the Management VLAN is tagged. You will need to make sure PS is aware of this.
During the preinstall meetings with the PS install team, they will cover some the network specific information for the VxRail install. The team will use that information to create a .json file which will contain the information on the following screens.
On the screen above is where you’d specify the NTP Server, but what I wanted to focus on was the NIC configuration. In my first post around vCenter Planning, there is a link to the VxRail Networking guide. With the E560, P570, V570 and S570 nodes, you have the option to deploy VxRail using either 2 of 4 of the 10GbE onboard ports on the NDC (network daughter card). If the last 2 onboard ports will be leveraged for VM Networks, you should deploy with the 2x10GbE option. Remember, VxRail lays out the networking configuration and you can only move VM Networks. The vSAN, Management and vMotion networks cannot be moved.
Another thing that PS will work through is the host naming scheme. It’s been a little while now, but we’ve added many options for naming the VxRail systems. Over the life of the VxRail, the options have continued to grow. With all shipping versions now, you can set the iterator, an offset and a suffix for the hostname. Early on, these were some of the feature requests we saw to help integrate VxRail into existing VMware environments. With the structure below, as long as the names have some sort of set hostname structure, VxRail should have you covered. The only thing we can’t do today is apply unique names to each node (think colors, cities, etc.). This is also where the range of IPs for the VxRail nodes are specified – you will need to provide a set of IPs that are contiguous and at least 1 per VxRail node. I would recommend that you plan accordingly and supply a range that’s larger, which will make node adds much easier.
In the planning post, I also had a pointer to the vCenter planning guide for VxRail. During the installation is where PS can tell the system to deploy an “internal vCenter” or tie into an existing vCenter Server in the environment. If the system is going to be installed into an existing vCenter, there is also an option to specify if the existing vCenter has an external PSC.
The next 3 screens help the system setup the vMotion, vSAN and VM Networks. For the vMotion, you can tie into an existing vMotion network as well. This is helpful in 2 scenarios 1) if the VxRail is being installed into an existing vCenter, it’ll make migrations easier and 2) if VxRail is being installed leveraged the internal vCenter, you can use this VMware fling to help with migration.
The “Solutions” screen allows the system to create a vRealize Log Insight instance to run on the cluster, if, and only if, the internal vCenter is leveraged. The license is included along with the internal vCenter license. If the system is deployed to an external vCenter, you can choose “none” or “syslog server” and point the VxRail to an existing Syslog Server within the environment.
The “Accounts” screen is where the passwords are set for the root accounts, as well as the management users are created on each of the nodes. There is an option to provide a shared password for some of the accounts, or you can choose to have a different password for each one, including a different root password for each ESXi host, should your security requirements dictate that.
The last step before actually deploying the system is to run the validation. This process takes the inputs on the previous screens and go through a validation of those values. During this process the system checks that the IPs aren’t in use, that forward and reverse lookup zones are created, NTP service is working and running, the switch is configured correctly by checking the VLANs along with some other tasks. Since the inception of VxRail, this is something that I’ve seen improve from every release. If the validation were to fail, the system will display an error message and once resolved, you can attempt the validate process again. If that validate succeeds, I would recommend you have the PS team click on the “Download JSON” button. This will allow you to have a JSON file with the variables used for the config.
Once the PS team clicks on “Build VxRail”, the first thing the system does is reset the VxRail Manager IP from the default (if the system is using it still – the default is changed earlier if the system is tying into an existing vCenter). Once that is completed, the PS team will click on “Start Configuration”.
At that point, the VxRail will go through the automation of numerous steps to create the VxRail vSAN based cluster. During this process, vCenter is deployed, the VDS is created and configured, vSAN Disk Groups are created, VMware HA is set up, EVC Mode is set, and many other tasks. When the install is complete (the time will vary based upon number of nodes, whether vCenter needs to be deployed, the number of disk groups and some other variables) the system, for all intents and purposes is ready for VMs.
Once this screen is displayed, the deployment is done. Also, please be sure to enjoy the fireworks and animation.