[lug] Setting up failover in Linux?

[Rob Nagler]

Hi Sean,

> Are you sure?  Because one of the links on that page is to their white-paper
> titled "Building Fault-Tolerant Applications on AWS Whitepaper"...

Did you read it?  It says things about how to make AWS fault tolerant,
but doesn't talk about building the apps themselves.

> But, I just don't know of anything in the open source arena that is going
> to achieve the goals you have laid out.

That's good input, thanks.

I have looked around for years, and no one seems to offer anything
like this.  I think the fact is that most systems fail in the small,
not in the large.  When S3 goes down due to a "bit flip", no one
really seems to care.   They even lost user data, and that didn't seem
to bother people (meaning that Amazon didn't lose any customers
afaict).  People seem to care more about cost than reliability.
Reminds me of those trucks with the signs "Our number one goal is
safety", and then I get cut off by the truck or its brakes are
smoking. :)  In trucks, I kind of get the advertising, but with web
services, I don't.  It's not a major cost to our business so why not
spend a bit more by going with a more reliable vender than AWS?

What we've got (mostly) is something that will not lose more than,
say, 5 minutes of user data.  Looking at application level fault
tolerance is quite interesting.  For example, Postgres has a nice log
copy mechanism, but the logs are always 16MB (compiled in constant) so
you can't run it every minute over the Internet without incurring way
more bandwidth than typical applications require.

FWIW, our solution will be (is already) OSS.

Thanks again for spending so much time to try to help.

Rob