This article will be the english version of my latest article about ESXi installation, configuration and applying best practises. Click here for the Turkish Version.
Applying best practices to your server virtualization platform will be very useful in terms of performance and troubleshooting as well as installing ESXi the right way. There are many things to be considered while installing and configuring ESXi but the following steps will be the important ones based on my experiences for people who have basic knowledge of ESXi installation.
1- Fimware and BIOS upgrade : Before installing any operating system to you physical server, it is recommended to upgrade firmware and BIOS versions.
For example; There is a tool named BOMC (Bootable Media Creator) for IBM (Lenova) servers in order to upgrade firmware and BIOS versions automatically. You can download latest updates with this tool and create a bootable iso file.
2- HCL check : You need to check the compability guide (Vmware HCL) in order to make sure if your hardware is compatible with the ESXi version before starting to install ESXi. If you install ESXi for test purposes and you have old hardwares you won’t probably find your hardware in the compability list but that does not mean you can not install ESXi. Therefore, you need to try to install ESXi to make sure whether it works or not.
3- Downloading ESXi : If your hardware is compatible with the intended ESXi version you can download ESXi from Vmware Web Site. At the same web page you will see Custom ISO files customized by hardware vendors. Those iso files include fixes, updates and drivers so you won’t need to spend time with checking HCL, installing drivers etc. I personally prefer to install custom iso files.
4- Installing ESXi : After downloading ESXi iso file you can boot your server with iso and begin installation process. There are many sources about installing ESXi such as this youtube video.
5- Network configuration : First thing to do after installing ESXi is to configure network settings. There must be at least 2 redundant physical ports for management network. It is better to configure low bandwidth ports for management and high bandwidth ports for virtual machine network.
It is needed to configure network infrastructure without “single point of failure” especially for enterprise organizations. It is essential to place redundant SAN and network switches in your infrastructure.
After installing ESXi with default configuration you can ping management IP address to make sure your network configuration is correct. If vCenter and hosts are working on a different subnet, you need to make sure if you have correct firewall rules and permissions.
After connecting your host, you will see one of uplink ports is stand-by adapter by default, even two uplink ports selected for management network. It is better to move upward the uplink port from stand-by adapters to active adapters in order to have uninterrupted redundancy.
6- Adding to dSwitch : If you have the correct network configuration you can now add your host to vCenter. You are now ready to add your host to dSwitch if you have an Enterprise Plus License.
You need to make sure that correct vmnics assigned as uplinks. There must be at least 2 uplinks assigned for redundancy.
Another important point is not to provide uplinks on the same network card because if there is a card failure you will be losing connections to all virtual machines working on that host. Therefore it is better to have two network cards with one single port instead of one network card with dual ports.
7 – Configuring distributed port group : Configuring port group with default settings is usually fine but i would like point out the load balancing setting. “Route based on originating virtual port” is selected by default as load balancing policy.
I prefer “Route Based on Physical NIC Load” as load balancing policy because it has some advantages such as;
– There is no need for physical switch configuration
– The distributed switch takes of the load of uplinks
– Minimal impact on resource consumption.
8- Configuring vmkernel adapters : After adding your host to an existing cluster, you need to configure vMotion network. When configuring port properties “vMotion” must be selected as TCP/IP stack.
If you intend to use NFS datastores there should be a specific vLan and vmkernel network for NFS traffic.
Note : I am planning to post another article about best practises for NFS datastore.
9- Syslog configuration : Syslog configuration should be done with one of existing datastores.
10- Power management : When you look into vmware best practises documents you will see that, “High Performance” should be selected as power policy setting especially for hosts running performance-centric virtual machines.
11- SAN configuration : SAN zoning configuration must be done after making physical connections between hosts and storages via SAN switches. Then hosts will have connections to datastores.
12- NTP configuration : One of the most important points is ntp configuration. Especially linux guest operating systems might be affected by the ntp configuration of the host and that causes failure of running applications.
Reference : I have tried to point out important steps of ESXi configuration based on my experinces. If you would like to get more details about performance best practices you can look into Vmware Reference Guide.