On Jun 10, 2014, at 1:18 PM, Robert Milkowski <rmilkow...@task.gda.pl> wrote:
>> -----Original Message----- >> From: Richard Elling via illumos-discuss >> [mailto:discuss@lists.illumos.org] >> Sent: 27 May 2014 14:42 >> To: gbul...@sonicle.com >> Cc: discuss@lists.illumos.org >> Subject: Re: [discuss] DLMP aggr >> >> >> On May 27, 2014, at 6:32 AM, Gabriele Bulfon <gbul...@sonicle.com> >> wrote: >> >>> Thanks for the clarifications. >>> >>> Do you mean I can just use ipmp over the 4 phys cards, have two of >>> them on switch 1 and two on switch 2, without switches lacp-trunking >> configuration, and I should get equivalent of dlmp? >> >> No. DLMP is just link failover, it is not aggregation. In DLMP only one >> physical link is in use at any time -- other links are standby. DLMP >> operates at layer 2. IPMP is similar, but operates at layer 3 (IP) and >> can be configured to do outbound load spreading. >> > > Actually, this is not entirely true. > If multiple vnics are created on top of dlmp then they will be spread across > the underlying interfaces. Yes, this is certianly a more complete answer, thanks Robert! Suppose we restate as: In DLMP only one physical link is used by a vnic at any given time. To take advantage of the bandwidth offered by multiple physical links, you will need to configure multiple vnics. -- richard > This will give you both in and out load spread in total, although each vnic > will be constrained by a single underlying interface. > > Also, with some switches one can do aggregation across switches as well. > > -- > Robert Milkowski > http://milek.blogspot.com > > > > > ------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com