Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 22 Mar 1999 11:45:59 -0500
From:      "Gary Palmer" <gpalmer@FreeBSD.ORG>
To:        "Cyril A. Vechera" <cyril@main.piter.net>
Cc:        julian@whistle.com, cmsedore@maxwell.syr.edu, freebsd-net@FreeBSD.ORG
Subject:   Re: clustering/load balancing 
Message-ID:  <60695.922121159@gjp.erols.com>
In-Reply-To: Your message of "Mon, 22 Mar 1999 11:07:20 %2B0300." <199903220807.LAA06347@main.piter.net> 

next in thread | previous in thread | raw e-mail | index | archive | help
"Cyril A. Vechera" wrote in message ID
<199903220807.LAA06347@main.piter.net>:
> 
> in the original scheme 'single point of failure' is still present.
> 
>                                   +-------[Machine B]
>                                   |
>  [internet]-----[ any router ]----+-------[Machine C]
>                                   |
>                                   +-------[Machine D]
> 		   ^^^^^^^^^
> 
> or maybe you can see other way to connect 'internet' to Machine [B-C]?

There are many ways of doing it... Just very few ways that remove the SPoF all 
the way to the host. HSRP'd routers and per-packet load balancing have reduced 
a lot of the router based failures that could happen, but that still doesn't 
remove the SPoF in the switch/hub between the routers and the hosts. As I said 
in a previous message, Alteon have failover between switches, although it 
probably requires their GigE cards to go to the host and have that work.

> what is the differense between 'any router' failures and 'balance
> dispatcher' failures?

I added a smiley to the end of the message for a reason. There are solutions 
to most of the other SPoFs that are out there today.  The biggest one still 
left is actually not the one I highlighted, but rather:


                                  +-------[Machine B]
                                  |
 [internet]-----[ any router ]----+-------[Machine C]
                                  |
                                  +-------[Machine D]
		              ^^^^^^^^^^^^

Anyone know of any way to have redudancy all the way to the host? (i.e. 2 or 
more NICs) Its going to need some daemon on the host watching the NIC for a 
heartbeat or something, then sending out an ARP invalidation packet for the 
(now failed) NIC and then another ARP for the (now working) NIC.

Gary
--
Gary Palmer                                          FreeBSD Core Team Member
FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info




To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?60695.922121159>