Mike Horwath wrote:
> On Fri, Dec 09, 2005 at 04:22:08PM -0500, Bob Dodds wrote:
>
>> Mike Horwath wrote:
>>
>>> On Dec 8, 2005, at 10:57 AM, Chris Ryland wrote:
>>>
>>>> This is a small point, but when I was worrying about MX backups, my
>>>> ISP's chief technical guy pointed out that it's not really necessary
>>>> these days--everyone out there is ready and able to deal with your
>>>> main mail host being down, and will queue locally.
>>>>
>>> As an operator of a largish ISP, I concur, there really isn't a need
>>> for a backup MX any longer.
>>>
>>> Some people like the fuzzy feeling, though, so I continue to operate
>>> one for those users, but there are only 8 or 10 domains now on it.
>>>
>>> Far better than the last ISP I ran - we had thousands of domains
>>> being backed up via MX, it was horrible and caused a lot of pain.
>>>
>> Is what was problematic that after the writing to disk, the resulting
>> complexities of subordinating the work queue to whether the primary
>> delivered messages successfully?
A cluster would have the same architectural constraints for
a work queue as a secondary or backup server setup. That's
why I ask. Where did the architecture fail? -Bob
Received on Mon Dec 12 22:52:07 2005
This archive was generated by hypermail 2.1.8 : Tue Dec 13 2005 - 00:00:01 EST