[openstack-dev] Neutron and MTU advertisements -- post newton

Jay Pipes jaypipes at gmail.com
Mon Jul 11 16:39:15 UTC 2016


On 07/11/2016 07:45 AM, Sam  Yaple wrote:
> Hello,
>
> There was alot of work to get MTU advertisement working well in Mitaka.
> With the work that was done we can finally have 1500 mtu networks for
> tunneled networks if the underlying infrastructure supports jumbo frames.
>
> Its fantastic for people who have 1500 mtu networks and want to use
> vxlan, no more hacks to get the instance to use 1450 mtu. Its fantastic
> for people who want to use 1500+ networks and get the instances setup
> with 9000 mtu interfaces. Its is not good for people who want consistent
> mtu no matter the network type. But thats fine, since mtu advertisement
> _could_ be disabled. Its a fantastic default to have it turned on.
>
> With a recent patchset [1] the ability to turn off MTU advertisements
> was deprecated in Newton. In the review it was stated there is no valid
> use case for it. I disagree.
>
> The scenario is the infrastructure has jumbo frames enabled, but I do
> not want the instances to be using jumbo frames, but I want them to be
> using the default 1500 mtu that the rest of the world operates on. This
> would still setup all of the virtual switching infrastructure to the
> correct MTUs, but not try to adjust the instances MTUs. In this way the
> instances are only communicating at 1500 mtu, but never having to
> fragment/drop inside of the SDN when communicating with other networks
> even if it is a VXLAN or other tunneled network.
>
> Without the option to disable mtu advertisement, inside the same
> environment flat/vlan and gre/vxlan network will always have different
> mtu, even if the backend supports jumbo frames.
>
> My ask is we keep the advertise_mtu option, and keep it enabled by
> default. This would allow for the default, common 1500 mtu across
> networks of different types.
>
> This scenario would be very similiar to having a computer with 1500 mtu
> attached to a switch which supports jumbo frames. Just because the
> switch will accept and process a 9000 mtu frame, doesnt mean the
> computer has to send a 9000 mtu frame. A very common scenario in the
> real world.
>
> [1] https://review.openstack.org/#/c/310448/

Hi Sam,

Out of curiosity, in what scenarios is it better to limit the instance's 
MTU to a value lower than that of the maximum path MTU of the 
infrastructure? In other words, if the infrastructure supports jumbo 
frames, why artificially limit an instance's MTU to 1500 if that 
instance could theoretically communicate with other instances in the 
same infrastructure via a higher MTU?

Sorry if my question is poorly worded. I'm no networking expert and am 
genuinely curious here. :)

Best,
-jay



More information about the OpenStack-dev mailing list