Redundant infrastructure

[Derek Atkins]

Hi,

Jon Daley <gnucash at jon.limedaley.com> writes:

> On Wed, 9 Aug 2017, Derek Atkins wrote:
>> All the solutions I know about are for multiple instances in the same
>> data center.  I have no idea how to do it in a globally distributed
>> manner.
>
> Failover can be setup via DNS, to ping/check a host and then swap IPs
> if the first host doesn't respond, and theb you can decide what you
> want to do when the first host comes back up, probably auto-switching
> back is harder to setup due to changes that happen on the second
> server while thw first is down.
>
> dnsmadeeasy.com has the cheapest (and actually best) service for this
> (as well as other things - I've used them commercially for ten years
> or so, and had almost no issues.

Right now Linas runs our DNS service.  I suspect we would have to do an
akamai-like www.gnucash.org CNAME www.gnucash.dnsmadeeasy.com or
some-such to get the load-balancing working.  I'm not sure I really want
a third-party out of our control to be able to repoint our services.

> But, for gnucash, I'd think you don't really need a high availability
> solution like this.  I'd expect good backups to be good enough,
> particularly if it is a virtual server that can be spun back up from
> the backup.  But, even real hardware is probably good enough - there
> probably aren't that many urgent requests that come that can't wait a
> day or two. And DNS could always be manually changed in the worst
> case.

Backups for www are easy -- the website is completely in GIT.
Backups for code are harder, but I have a nightly backup to another
system, but it's still local to my network.
Neither of these are "decent" for disaster recovery.

> All that said, if you want to go that route, I have experience with
> this and can spend time or at least recommend software options.

I think we would first need to decide what we actually want/need, and
then we can look at what we have and determine what steps will be
necessary to take what we have and get what we want/need.

I don't think we've fully analyzed the first step, yet, even though
we're talking about the second.

-derek

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord at MIT.EDU                        PGP key available