vSwitch failover question

[JasonVmware]

Hello All,

I was wondering when you setup a vSwitch for you service console / vmkernal should you go into the vSwitch properties and change the load balancing properties for that vSwitch to Use Explicit Failover only, or just leave it as Route based on the orginating port id?

On my vSwitch for my service console and vmerkernal / vmotion I left it as Route based on the orginating port id and just set the failover on my port groups, however I was wondering if I should change the default setting for this vSwitch or not.

[Troy_Clavell · Accepted solution]

we don't change our vSwitch load balancing from the default which is what you are using. However for a Service Console and Vmotion port groups, both on vSwitch0, we load balance the pNICs to be standby for each other.

if you feel this or any other answer was useful, please consider marking it as "correct" or "helpful"

[Ken_Cline]

If you have a single vSwitch with two affiliated pNICs and only two internal "interfaces" connected, you don't need to specify failover order. The default behavior will be to affiliate pNICs based on vSwitch port. Since there are only two connections to the vSwitch, each will - by default - affiliate with a different pNIC. The only "downside" to letting everything default is that you don't know, deterministically, which interface will be active for what service. The question is - do you care?

Ken Cline

Technical Director, Virtualization

Wells Landers

TVAR Solutions, A Wells Landers Group Company

VMware Communities User Moderator

[TheCandyMan]

One of the questions I asked my self severa times (And therefore ask you on this occassion) is: why shoudn't I just use a pNICS as active a the time ? Eventually, they will fail over for each other in case of a failure. So why no utilizing the added bandwith on the ESX and keep both NICs active ? Why keeping one "IDLE" ?

Thx.

[Ken_Cline]

why shoudn't I just use a pNICS as active a the time ? Eventually, they will fail over for each other in case of a failure. So why no utilizing the added bandwith on the ESX and keep both NICs active ? Why keeping one "IDLE" ?

I tend to agree with you. The reasons I've had customers give for setting standby pNICs are:

- - Deterministic network behavior. They want to know what traffic is using which interface.

- - "Legacy" pSwitches. They're in the process of upgrading their network, but aren't done yet.

- - They're "tinkerers" and they just have to adjust every setting they can

- - They don't really understand how ESX networking works

In general, my rule of thumb is this: if you don't have a good reason to change a default behavior, don't change it

Ken Cline

Technical Director, Virtualization

Wells Landers

TVAR Solutions, A Wells Landers Group Company

VMware Communities User Moderator

[TheCandyMan]

So basically, I can define these people as:

1. They are probably the "Paper MCSE" type... Not the ones who want to understand HOW and WHY it works as it is...

2. They Are control freaks

3. They are probably people who work with Robot-like "Server Hardening Guides" without really understanding what exactly they have or need to "Harden"

3. Just for spite...

:smileydevil:

[Troy_Clavell]

I tend to not quite understand the reasoning for the active standby NICs myself. We have multiple vCenter instances some clusters consist of the active/standy and some have both active. Though the active/standy was told to us by VMware Engineers and Consultants as best practice, they haven't really explained to us the reasoning behind this.

In my experience both setups work great, I think it's just based on how you want your environment setup. Having two NICs active doesn't guarantee faster speeds if you are connected to a 1GB backend.

Good stuff guys!!