Instead of using flow stats per NUMA node, use it per CPU. When using megaflows, the stats lock can be a bottleneck in scalability. On a E5-2690 12-core system, usual throughput went from ~4Mpps to ~15Mpps when forwarding between two 40GbE ports with a single flow configured on the datapath. This has been tested on a system with possible CPUs 0-7,16-23. After module removal, there were no corruption on the slab cache. Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@redhat.com> Cc: pravin shelar <pshelar@ovn.org> Acked-by: Pravin B Shelar <pshelar@ovn.org> Signed-off-by: David S. Miller <davem@davemloft.net> |
||
---|---|---|
.. | ||
Kconfig | ||
Makefile | ||
actions.c | ||
conntrack.c | ||
conntrack.h | ||
datapath.c | ||
datapath.h | ||
dp_notify.c | ||
flow.c | ||
flow.h | ||
flow_netlink.c | ||
flow_netlink.h | ||
flow_table.c | ||
flow_table.h | ||
vport-geneve.c | ||
vport-gre.c | ||
vport-internal_dev.c | ||
vport-internal_dev.h | ||
vport-netdev.c | ||
vport-netdev.h | ||
vport-vxlan.c | ||
vport.c | ||
vport.h |