Jamie Honan wrote:
> > Mail service outages can only happen to the grossly incompetent, the
> > disinterested
>
> > Major ISPs have no excuse for mail service outages at all.
>
> Ouch.
>
> Telstra should make their own excuses, but this is a very hard
> one.
>
> Maybe what you assert would be possible with a lot, lot more
> expenditure by ISP's. I would never ascribe it to incompetence.
It is not expensive, this is a (very) common misconception. As the
service gets larger, the opportunites for introducing no cost (or
minimal cost) redundancy actually increase. Telstra, however, appears to
have decided to ignore decades of development in best practice for very
large systems and has done this:
$ mx bigpond.com.au
bigpond.com.au MX 10 extmail.bigpond.com
$ mx bigpond.com
bigpond.com MX 10 extmail.bigpond.com
$ mx telstra.com
telstra.com MX 5 extmail.bigpond.com
$ host extmail.bigpond.com
extmail.bigpond.com A 139.134.5.153
This is a disaster waiting to happen. This I ascribe to incompetence.
(Notes:
- For those who don't follow, the problem is that having a single MX
means that if that MX is down, the domain's entire mail service is down.
- Telstra is not the only large ISP that exposes itself this way. I am
not suggesting that Telstra is the world's worst, but that it's up there
with them.
- At least someone inside Telstra understands the risks associated with
the lack of redundancy: telstra.com.au has two MXs.
- It is likely that 139.134.5.153 has something comparable to a Cisco
Director on it, and a farm of actual mail servers behind it. This is not
enough, the Director and the network leading to it remain single points
of failure.
)
- Raz
--
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://slug.org.au/lists/listinfo/slug
