Replication Scene Consultant

Hi Guys!
I work in Govern Department Bank. We just have many experiences about
Replication, meanly Transactional.
All evolution about Client-Server architecture in the past, make some choice:
?Many Distribution application at many sites located in many cities linked
through Wan (frame relay – 98 kbps to 512 kbps)
?Each site has one Domain controller Server, with SQL 2000 Standard, 2
processor and 1 Gbyte of Memory. Many Databases that belongs each other to
your own application and share some tables.
Now we need put all information together at central Server. Let’s take a look:
?We have many databases : Ddes950, Ddes033, etc
?Many tables that become at each other Database needs to plus at one table,
I mean, if one table at site Ag002,Ag003, Ag004,…, call T999xxx01, we need to
add to one central. The distributed table has different PK row at each other!
Transactional replication can do this administration?
How can I do that without truncate central table after every snapshot
initialize schema?
How can I control synchronisms and snapshots ?
How can I control schema changes?
Could Microsoft replication support many publication to one subscriber?
What about Oracle solutions? Can we plus together with Microsoft SQL Server?
Please, if anyone has any information or Scene like that situation, please
send me and share these with us!!!
Thanks
Krisnamourt
Message posted via http://www.droptable.com
It sounds like a multiple publishers and central subscriber topology is what
you need. Here is an article on setting it up etc :
http://www.replicationanswers.com/CentralSubscriberArticle.asp
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com .
|||Paul,
It's more complicated!
I follow that article example, and make some experiences like that, but here
I have more to worry:
The total table is something like 100.
The table at publisher just have many rows, I mean, some table have 3
millions rows or more, so the Initial SnapShot will required at firts charge,
so How can I coordinate this ? That's the point!!!
I do another experience with snapshot files manipulations, so the syncronism
isn't the problem, but Initial snapshot schema is...
Sds!
Paul Ibison wrote:
>It sounds like a multiple publishers and central subscriber topology is what
>you need. Here is an article on setting it up etc :
>http://www.replicationanswers.com/CentralSubscriberArticle.asp
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via http://www.droptable.com
|||Paul,
It's more complicated!
I follow that article example, and make some experiences like that, but here
I have more to worry:
The total table is something like 100.
The table at publisher just have many rows, I mean, some table have 3
millions rows or more, so the Initial SnapShot will required at firts charge,
so How can I coordinate this ? That's the point!!!
I do another experience with snapshot files manipulations, so the syncronism
isn't the problem, but Initial snapshot schema is...
Sds!
Paul Ibison wrote:
>It sounds like a multiple publishers and central subscriber topology is what
>you need. Here is an article on setting it up etc :
>http://www.replicationanswers.com/CentralSubscriberArticle.asp
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via http://www.droptable.com
|||Paul,
It's more complicated!
I follow that article example, and make some experiences like that, but here
I have more to worry:
The total table is something like 100.
The table at publisher just have many rows, I mean, some table have 3
millions rows or more, so the Initial SnapShot will required at firts charge,
so How can I coordinate this ? That's the point!!!
I do another experience with snapshot files manipulations, so the syncronism
isn't the problem, but Initial snapshot schema is...
Sds!
Paul Ibison wrote:
>It sounds like a multiple publishers and central subscriber topology is what
>you need. Here is an article on setting it up etc :
>http://www.replicationanswers.com/CentralSubscriberArticle.asp
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via droptable.com
http://www.droptable.com/Uwe/Forums.aspx/sql-server-replication/200612/1
|||I'm not too sure which part of your setup is more complicated than that in
the article.
You have a lot of tables and a lot of records but the methodology should be
exactly the same, assuming the data is all partitioned. Are you saying that
the data isn't partitioned ie there is some overlap of data across the
various publishers?
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com .
|||Let me explain more:
When I create Publisher and subscriber with “keep the existing table
unchanged”, we need initial load too.
When I maked for one publisher ok, but when another publisher make the same,
It don′t plus row one by one, It truncate the table and plus its own snapshot.
THAT’S THE POINT!
Every time , the process make the same!
What's the magic? Please , tell me
I would need to do the load process in separated way?, I mean:
1.I’ll create the publish and subscriber , without any load.
2.Stop the distribution agent and let log reader agent run
3.Start and finish the load by DTS or BCP.
4.Start distribution agent again. Pretty !!!
If I’ll do in this way, work fine, but It’s more hard to do and to manage.
Sds!
Kris
Paul Ibison wrote:
>I'm not too sure which part of your setup is more complicated than that in
>the article.
>You have a lot of tables and a lot of records but the methodology should be
>exactly the same, assuming the data is all partitioned. Are you saying that
>the data isn't partitioned ie there is some overlap of data across the
>various publishers?
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via http://www.droptable.com
|||Let me explain more:
When I create Publisher and subscriber with “keep the existing table
unchanged”, we need initial load too.
When I maked for one publisher ok, but when another publisher make the same,
It don′t plus row one by one, It truncate the table and plus its own snapshot.
THAT’S THE POINT!
Every time , the process make the same!
What's the magic? Please , tell me
I would need to do the load process in separated way?, I mean:
1.I’ll create the publish and subscriber , without any load.
2.Stop the distribution agent and let log reader agent run
3.Start and finish the load by DTS or BCP.
4.Start distribution agent again. Pretty !!!
If I’ll do in this way, work fine, but It’s more hard to do and to manage.
Sds!
Kris
Paul Ibison wrote:
>I'm not too sure which part of your setup is more complicated than that in
>the article.
>You have a lot of tables and a lot of records but the methodology should be
>exactly the same, assuming the data is all partitioned. Are you saying that
>the data isn't partitioned ie there is some overlap of data across the
>various publishers?
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via droptable.com
http://www.droptable.com/Uwe/Forums.aspx/sql-server-replication/200612/1
|||Let me explain more:
When I create Publisher and subscriber with “keep the existing table
unchanged”, we need initial load too.
When I maked for one publisher ok, but when another publisher make the same,
It don′t plus row one by one, It truncate the table and plus its own snapshot.
THAT’S THE POINT!
Every time , the process make the same!
What's the magic? Please , tell me
I would need to do the load process in separated way?, I mean:
1.I’ll create the publish and subscriber , without any load.
2.Stop the distribution agent and let log reader agent run
3.Start and finish the load by DTS or BCP.
4.Start distribution agent again. Pretty !!!
If I’ll do in this way, work fine, but It’s more hard to do and to manage.
Sds!
Kris
Paul Ibison wrote:
>I'm not too sure which part of your setup is more complicated than that in
>the article.
>You have a lot of tables and a lot of records but the methodology should be
>exactly the same, assuming the data is all partitioned. Are you saying that
>the data isn't partitioned ie there is some overlap of data across the
>various publishers?
> Cheers,
> Paul Ibison SQL Server MVP, www.replicationanswers.com .
Message posted via droptable.com
http://www.droptable.com/Uwe/Forums.aspx/sql-server-replication/200612/1
|||I still think something is set up incorrectly as this can work fine. The
initial publication has article properties which are the default ie it uses
a 'drop table' as the initial script. The subsequent publications use the
option to "keep the existing table unchanged". Run the snapshots for these
subsequent publications without subscribing to test - there should be no
drop statement or truncate or delete statements in the snapshot folder and
the data should still be in BCP data files like usual. If you have something
different to this (ie a drop etc) then please double-check the article
properties as something is seriously awry. If you have an article set up
with "keep the existing table unchanged" and yet there is a
drop/truncate/delete, please script up your publication and table and I'd
like to reproduce it.
Cheers,
Paul Ibison SQL Server MVP, www.replicationanswers.com .