LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: The Two Server Distance Problem (Re: DNS Trick??)

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: The Two Server Distance Problem (Re: DNS Trick??)
From: Roberto Nibali <ratz@xxxxxx>
Date: Wed, 30 Apr 2003 17:06:22 +0200
Hi Joel,

Maybe I am just ingorant with LVS.

I hope not.

What I need is redundancy between just two computers that is on the internet at two different locations.

I assumed you already have redundancy in your setup so that's why I insisted on getting the information about where and how this redundancy is happening. The point is, that if you control the redundant point, you should put the LVS box there.

The DNS has one entry for the VIP which then gets forwarded over a tunnel to 
the RS.

I was thinking of doing a cluster with load balancing, but then I would have to have a router at one location to balance it which is a failing point.

Why?

Is there any way to have the load balancers running on both machines that the web services are being published??

Yes, but it doesn't help the problem. The root of the problem is where your traffic split is done to the geographical endroits. There you must provide a redundancy box with healthchecking and service takeout.

The only problem that arises with this scenario is if the website is using a database. I don't know if the MySQL daemons would be able to handle the

Do you plan on using one or is it already in place?

traffic with the databases not being in close proximity to each other. And all of the database queries that are getting written to them would need to be secure as well.

To my knowledge, it would be extremely hard to have SQL DBs synced over distance. But you can write your web application in a way that will upgrade the other DB too (remote DB) on a request. Then you will need persistency and transaction safety.

If one location fails, all customers will be redirected to the other location, DB will get new entries, which will be required to be sync'd over to the failed location once it is up again. Depending on the amount of new data coming into your DB you can do it with a couple hundreds of lines of shell/perl code. If I find something I'll let you know.

Cheers,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc

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