Date: Tue, 29 Jun 2021 08:28:41 +0200 From: Lutz Donnerhacke <lutz@donnerhacke.de> To: Rick Macklem <rmacklem@uoguelph.ca> Cc: freebsd-net <freebsd-net@freebsd.org> Subject: Re: RFC: NFS trunking (multiple TCP connections for a mount Message-ID: <20210629062841.GC31225@belenus.iks-jena.de> In-Reply-To: <YQXPR0101MB0968DC173855A82AAF45F08FDD039@YQXPR0101MB0968.CANPRD01.PROD.OUTLOOK.COM> References: <YQXPR0101MB0968DC173855A82AAF45F08FDD039@YQXPR0101MB0968.CANPRD01.PROD.OUTLOOK.COM>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Jun 29, 2021 at 12:23:21AM +0000, Rick Macklem wrote: > I don't understand how multiple TCP connections to the same > server IP address will distribute the load across multiple network > interfaces? Sounds like a bandwith-latency-product issue. TCP is prone to stalling at the end of a buffer window (in addition to reassembly and RPC waits). Multiple TCP streams scale linear with the available bandwidth over them. Single RPC messages should not split over multiple streams (causes reorder of segments), but different messages can go in round robin fashion. There is no need to ensure a symmetric path, so that the response comes back on the same stream.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20210629062841.GC31225>