January 23, 2003 at 12:07 am
hi...i was instructed by my boss to do some research on one replication scenario and come out with a solution.
The scenario is.. theres a main db (big one) will push data to over 60 small db and the system on normal basis will push only the differences to those small db. Plus theres an option where the system can do drop and re-create tables (snapshot replication).
I have been looking at DTS, Replication Wizard and Trans-SQL ...do know which ones to choose...all of them got pros and cons..and performance is very important..please help...
thanks
January 23, 2003 at 4:40 am
If you are wanting to push only changes then snapshot is out but it will be used on the initial push either way. So you are left with merge (which adds a guid field to the data) or Transactional which monitors the Tranbsaction Log and copies over the transaction. Those will be you considerations for replication. Then of course in DTS you will need to flag the changes somehow to know to push down to the other DBs so that means control tables or columns of some kind need to be developed.
Now to be able to give you some idea as to what can you walk us thru what each of the small dbs contain and what type of flow you are actually after?
January 23, 2003 at 11:42 am
I do something similiar in reverse using transaction replication an find it fairly straight forward. I have about 15 servers replicating data to a central site.
I use Pull vs. puch to aleviate the load on the remote servers. There are some problems but it works well.
I think the biggest problem you will have, which may require push, is to filter the data for each location. Or can you just ship all changes / data to all satelites ?
KlK, MCSE
KlK
January 23, 2003 at 7:11 pm
Hi...
Thanks for the replies....
Ok...actualy this scenario is in a manufacturing factory...currently there main db that control all the logics of the production line...because of so many trafic to this main db..my boss came out with a sugguestion to split the main db into smaller db to relief the main db from network trafic...this small db will be put at every production line..so it can be more than 60 small db....so i was thinking do a straight forward transactional replication from main db to those small db...because every changes made in main db will replicated to those small db..so im thinking to make a custom app or scripts that support custom subscribers and articles...so im looking at straight forward transctional replication wizard at the enterprise manager or do i have to develop custome dts/scripts to help solve this problems.. hope u guys can help me..
January 23, 2003 at 7:37 pm
hi..
for ur info...actualy there are 2 main db one for the business logic and another for transactional data and 1 line can have max 9 stations and min 6 stations. plus average this stations hits the main db for business logic are 15 times...and there are more 5000 transaction per day.. right now average production speed for 1 product is 30 seconds from first station to the last station.. so my main focus is to minimized network trafic to the business logic db and speed up the production line...our target is to speed up to 20 secodes per product...thats why main boss came out with this suggestion..
Edited by - amirx on 01/23/2003 7:40:57 PM
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply