Hi there,
According to VMware Knowledge Base to enable jumbo frames (mtu 9000) on a vmkernel, you need to enable it at the vmkernel level.
I was wondering, if I let the vDS default MTU of 1500, but enable MTU 9000 on a specific vmkernel port, will this work ?
As jumbo frames requires end-to-end configuration, if the vds is not configured for jumbo frames, I think it should not work.
But the doc is unclear about that.
Also, I think I have the answer, but just to be sure :
With standard vSwitch, teaming and failover configuration is done at vSwitch level.
Seems like it's not possible to do the same at vds level and you need to configure this for EACH portgroup... Anyone confirm ?
Thanks folks
AFAIK with vDS, these policies are defined at the portgroup level. Then, they can be overridden at the individual ports level. Also, either all the portgroups can be similarly configured for the policies or they can have their individual configuration. More details in the below links -
Manage Policies for Multiple Port Groups on a vSphere Distributed Switch
Configure Overriding Networking Policies on Port Level
Cheers,
Supreet
As per my understanding, Jumbo Frames has to be enabled on both switch (standard and distributed) and the required vmkernel ports. Having vSS/vDS at 1500 and the vmkernel port at 9000 or vice versa will not allow Jumbo Frames through the vmkernel port in question. Personally, I have seen such instances wherein there is a mismatch between the vSS/vDS and the vmkernel ports. Be it vSS or vDS, we will have an override option at the portgroup level. So the teaming and failover options can be configured specific to the portgroup as well.
Cheers,
Supreet
Jumbo Frames must be configured correctly end-to-end. This means on the vSS/vDS, vmkernel interfaces, physical switches, endpoint everything must be configured to support the larger frames end-to-end. Using larger frames will provide more efficient processing of network traffic but if any part of the path is mis-configured, network frames larger than 1500 may not pass.
vmkping command can be used to test a vmkernel interface on the ESXi host. This is very useful for testing proper configuration of Jumbo Frames.
Hi Sharantyr3
Changing MTU on vmKernel port only is not enough. MTU should be changed on VDS and on the physical network path all the way as well. Otherwise, fragmentation will occur.
For more information, Please check my below blog about changing MTU in vSphere environment and how to test it via CLI.
MTU Change on a vSphere Environment - vExpert Consultancy
Hi,
Like I thought yes, but the official KB is not clear about that.
Also, about my second question, any idea ?
When you create a new PG on dvs, it offers you a checkbox "customize default policies configuration".
When you check this box, you get the "traffic shaping", "teaming and failover", etc options.
These options seems not changeable at the dvs level whereas you can change it on a standard vswitch level.
Or maybe via CLI ?
AFAIK with vDS, these policies are defined at the portgroup level. Then, they can be overridden at the individual ports level. Also, either all the portgroups can be similarly configured for the policies or they can have their individual configuration. More details in the below links -
Manage Policies for Multiple Port Groups on a vSphere Distributed Switch
Configure Overriding Networking Policies on Port Level
Cheers,
Supreet