LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: [lvs-users] LVS in an Active/Active configuration

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: [lvs-users] LVS in an Active/Active configuration
From: "Christopher Barry" <christopher.barry@xxxxxxxxxx>
Date: Thu, 3 Jan 2008 08:19:51 -0600


-----Original Message-----
From: lvs-users-bounces@xxxxxxxxxxxxxxxxxxxxxx on behalf of Graeme Fowler
Sent: Thu 1/3/2008 6:18 AM
To: LinuxVirtualServer.org users mailing list.
Subject: Re: [lvs-users] LVS in an Active/Active configuration
 
>Hi

>On Wed, 2008-01-02 at 13:38 -0500, Matthew Crocker wrote:
><snip excellent topology description>

>Aha, equal cost paths using OSPF. Neat, I hadn't considered that - nice
>to see someone making proper use of L3 "trickery" for once. And it
>completely removes the necessity for your upstream network devices to
>make use of ARP, which simplifies things even more. The removal of the
>"live" IP addresses from your directors' external interfaces makes
>things even simpler.
>
>Why didn't I think of that? ;-)

>> LVS-NAT won't work because of the requirement that returning traffic  
>> has to pass through the correct director.  LVS-DR would probably work  
>> fine, the real servers could then send the return traffic directly  
>> back to the routers.

>As I previously mentioned, you could split your realservers into two
>groups; or you could take Chris Barry's suggestion of running the sync
>daemon in both master and slave state on both directors to keep the
>tables bang up to date (inasmuch as that's possible).

>Using DR negates that level of complexity - and because you're routing,
>rather than switching, traffic to the directors you don't need to solve
>any ARP problems.

>>From here, Matt, this looks like a very good solution.

>Graeme



Hmmm. Being a victim of your own genius, Matt, it looks like someone has
a HowTo to write...

;)




<<winmail.dat>>





<Prev in Thread] Current Thread [Next in Thread>