Re: Call for 7.5 feature completion

Hans-Jürgen Schönig wrote:

> Marc G. Fournier wrote:
>> On Mon, 17 May 2004, Bruce Momjian wrote:
>>
>>
>>>Marc G. Fournier wrote:
>>>
>>>>>Agreed, but you are a "me too", not a huge percentage of our userbase.
>>>>
>>>>How do you know? Have you polled our complete userbase?
>>>>
>>>>
>>>>>Basically, after 6-7 months of development, I want more than a vacuum
>>>>>patch and a new cache replacement policy. I want something big, in
>>>>>fact, several big things.
>>>>
>>>>Most likely won't happen, since what is considered big by you isn't
>>>>necessarily what is considered big by someone else ... as Hannu, and I
>>>>believe, Jan, have so far pointed out to you ...
>>>
>>>I can't poll for everything. I make my own educated guesses.
>>
>>
>> Based on what though?
>>
>> All the clients that I deal with on a daily basis generally care about is
>> performance ... that is generally what they upgrade for ... so, my
>> 'educated guess' based on real world users is that Win32, PITR and nested
>> transactions are not important ... tablespaces, I have one client that has
>> asked about something *similar* to it, but tablespaces, for him, doesn't
>> come close to what they would like to see ...
>>
>> So, my 'educated guess' is different then yours is ... does that make
>> yours wrong? Nope ... just means we have different sample sets to work
>> with ...
>>
>
>
> Interesting.
> We have made COMPLETELY different experiences.
>
> There is one question people ask me daily: "When can we have sychronous
> replication and PITR?".
> Performance is not a problem here. People are more interested in
> stability and "enterprise" features such as those I have mentioned above.
>
> I am still wondering about two things:
> Somebody has posted a 2PC patch - I haven't seen too many comments
> Somebody has posted sync multimaster replication (PgCluster) - nobody
> has commented on that. Maybe I am the only one who has ever tried it ...

Do you really need someone "commenting" on query based replication? I
get goosebumps from just thinking someone would voluntarily push all
sequence- or timestamp-generation and other not strictly deterministic
functionality into the application to be able to use such a "solution".
This is exactly how people work around all the MySQL idiosyncrasies.

>
> Most likely this is not very encourageing for the developers involved ...

Most hopefully this is very discouraging! Connection pools are a nice
thing and I have used pgpool recently with great success, for pooling
connections. But attempting to deliver multimaster replication as a
byproduct of a connection pool isn't going to become an enterprise
feature. And the more half-baked, half-functional and half-reliable
replication attempts there are, the harder it will be to finally get a
real solution being recognized.

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #