[ovs-dev] [PATCH net-next 1/6] netdev-dpdk: Allow vswitchd to parse devargs as dpdk-bond args
i.maximets at samsung.com
Thu Apr 12 14:46:26 UTC 2018
On 12.04.2018 17:18, Jan Scheurich wrote:
>>> The bond of openvswitch has not good performance.
>> Any examples?
> For example, balance-tcp bond mode for L34 load sharing still requires a recirculation after dp_hash.
dp_hash is lightweight action now since we're using rss hash for it.
Recirculation itself is not a very big issue. Last time we measured
DPDK bonding in XOR balancing mode with OVS, it had more than 30%
of performance overhead in compare with simple DPDK port in a L2
"receive -- > send-back" scenario and a single PMD thread.
> I believe that it would definitely be interesting to compare bond performance between DPDK bonding and OVS bonding with DPDK datapath for various bond modes and traffic patterns.
Maybe, but I would not expect much profit from using DPDK bonding.
> Another interesting performance metric would be link failover times and packet drop (at link down and link up) in static and dynamic (LACP) bond configurations. That is an area where we have repeatedly seen problems with OVS bonding.
>>> In some
>>> cases we would recommend that you use Linux bonds instead
>>> of Open vSwitch bonds. In userspace datapath, we wants use
>>> bond to improve bandwidth. The DPDK has implemented it as lib.
>> You could use OVS bonding for userspace datapath and it has
>> good performance, especially after TX batching patch-set.
>> DPDK bonding has a variety of limitations like the requirement
>> to call rte_eth_tx_burst and rte_eth_rx_burst with intervals
>> period of less than 100ms for link aggregation modes.
>> OVS could not assure that.
> A periodic dummy tx burst every 100 ms is something that could easily be added to dpif-netdev PMD for bonded dpdk netdevs.
I don't think that it could be done easily, because port could be never used
for sending, or even not polling by any PMD thread.
If these dummy operations will be called from a separate thread we'll have
to protect everything by a mutexes. Even harder will be ensuring of required
Anyway this will be one more dirty crutch in a generic code.
I'd prefer not to do that.
>>> These patches base DPDK bond to implement the dpdk-bond
>>> device as a vswitchd interface.
>>> If users set the interface options with multi-pci or device names
>>> with ',' as a separator, we try to parse it as dpdk-bond args.
>>> For example, set an interface as:
>>> ovs-vsctl add-port br0 dpdk0 -- \
>>> set Interface dpdk0 type=dpdk \
>>> And now these patch support to set bond mode, such as round
>>> robin, active_backup and balance and so on. Later some features
>>> of bond will be supported.
>> Hmm, but you're already have ability to add any virtual dpdk device
>> including bond devices like this:
>> ovs-vsctl add-port br0 bond0 -- \
>> set Interface dpdk0 type=dpdk \
>> So, what is the profit of this patch-set?
> Thanks for the pointer. That is a valid question.
> I guess special handling like periodic dummy tx burst might have to be enabled based on dpdk-devargs bond configuration.
> BR, Jan
More information about the dev