[ovs-dev] [PATCH 1/1] dpif-netdev: Conditional EMC insert

[Jan Scheurich]

> It's not a big deal, since the most important use case we have for
> dpif-netdev is with dpdk, but I'd still like the code to behave
> similarly on different platforms.  How about defining a function that
> uses random_uint32 when compiling without DPDK?
>
> For testing it's not that simple, because unit tests can be run with
> or without DPDK.  It would need to be configurable at runtime.
> Perhaps making EM_FLOW_INSERT_PROB configurable at runtime would also
> help people that want to experiment with different values, even
> though, based on the comments, I guess they wouldn't really see much
> difference.
>
> Again, what do you think about simply using counting the packets and
> inserting only 1 every EM_FLOW_INSERT_PROB?
>
> Thanks,
>
> Daniele

As far as I know Ciara did some quick tests with a counter-based 
implementation and it performed 5% worse for 1K and 4K flows than then 
current patch. Perhaps we could find the reason for that and fix it, but 
I also feel uncomfortable with deterministic insertion of every Nth 
flow. This could lead to very strange lock-step phenomena with typical 
artificial test work loads, which often generate flows round-robin. I 
would rather use a random function, as you suggest, or count "cycles" 
differently when compiling without DPDK.

I agree to making the parameter EM_FLOW_INSERT_PROB configurable for 
unit test or other purposes. Should it be a new option in the 
OpenvSwitch table in OVSDB or rather a run-time parameter to be changed 
with ovs-appctl?

Jan