Am 17.09.16 um 00:40 schrieb Davide Poletto:

Hope not be wrong here but Port Trunking usage deserves its part in the whole picture too: be aware that using Port Trunking (with LACP as per IEEE 802.3ad) between your Servers' NIC and your 10Gb Switching infrastructure - and this happens by aggregating "n" identical ports together on both link's ends, as you wrote - doesn't consequently mean that your "one-Host-to-one-Host" generated traffic will be able to use and will be able to saturate all those "n" 10Gb based links concurrently...


It's a common misunderstanding that bundling a couple of ports together with LACP results in bigger throughput, as one connection will always only deliver it's maximum physical bandwidth. There are bonding modes, which support that, but you'd have to be able to configure this end to end, with some round-robin settings, as of on wich NIC the next packet will be stuffed into the "pipe". Regular 802.3ad will not give you that, as it has been designed for redundancy, not performance.

So, LACP gives you mutiple read streams up to the number of NICs involved in the bundle, but writes from the host will always be on one NIC for any given connection. The classic algorithm is based on the source and target MAC addresses. So you will not be able to get 2 to 3 GB/s over a 10GbE LACP bundle.

OmniOS-discuss mailing list

Reply via email to