June 28, 2004 at 8:28 am
Hi
I have several (identical) real time systems feeding data out to local (identical) databases. I need to merge all data to a central (again identical) database. The bottleneck is expected at the client, so merge replication should place no performance burden. Every few minutes (~5), i need to MOVE data from a (95% connected, so not 24/7) client to the central database. transaction histories are not required.
Two questions:
(1) would merge replication be my best option? I consider DTS, but datbase integrity (each database cotnains multiple, connected, tables) is an issue.
(2) is it possible to MOVE data from clients to server, as opposed to replicating the ata? When data is deleted from the client, it should not be removed from the server!
Kind regards
j
June 28, 2004 at 5:25 pm
June 29, 2004 at 1:31 am
hi
every few seconds, assuming a network connecrion is available.
given the multiple RT systems, i'd prefer to cycle through them - gathering data in an effiecent way, and getting the data to the central server within a minute of production.
Question - you say merge replication would be fine - but are there ways to return as delete, (i.e. to delete rows once moved to the server), etc?
Cheers
j
June 29, 2004 at 4:29 pm
Given the fact that you need to delete messages from the client machine you are forced to use a DTS package running at intervals. The Replication models are just that - Replication of data from one server to another. Therefore if you delete data at one end the other end will mirror this exactly.
Once you get into a model of needed disparate data then you need to move to DTS or a combination of Replication and DTS.
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply