[erlang-questions] Fault-Tolerant TCP/IP Servers

[Jon Gretar Borgthorsson]

Usually people do not do this on a software level. It's highly unusual and
I'm going to be brave and state that it's completely the incorrect way do do
this. It just simply increases the possibility of major security issues when
you start giving your programs full access to system resources. It's not an
erlang thing. You really should not do this in any programming language.
You have 2 choices.

Use a High Availability tool like Linux-HA. In an Active/Passive environment
you usually have the computers with their own ip address but then another IP
address that can freely travel between the computer as their secondary
address.

Or you can set up an Active/Active system that allows you not only fault
tolerance but load balancing as well. The downside to that is that you need
the third computer as a load balancer or preferably a load balancing
appliance.

And I highly recommend talking to a networking expert. It's no use being the
world best programmer if the whole thing fails then on the network level.


2008/7/16 David Mercer <dmercer@REDACTED>:

>  I must admit, I am disappointed that there is no Erlang solution.  This
> seems like a basic building block to fault-tolerant systems.  So, I am going
> to toy with my idea of having the secondary detect the failure and send the
> appropriate commands to the network to redirect traffic for the primary
> server's IP address to the secondary.  Am thinking there must be some DHCP
> commands to reassign the IP address.  Before I begin, does anyone see any
> obvious flaws in my thinking.  I am not a networking expert.  Thanks.
>
>
>
> David
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20080716/3830f168/attachment.htm>