May 8, 2008 at 11:43 am
Hi,
I'm pretty much new to SQL replication and not sure that I'm fully aware of all its aspects. Any help you can give me on my questions would be greatly appreciated.
Server A (SQL 2005) is at the central office. Server B (SQL 2000) is half way around the globe.
On a regular basis (every 15-30 min), we need to replicate data over to the central office from the remote regional office. The data at the remote office is just a small portion (subset) of data at the central office ... regional sales, RMA, product info, accounting data (which needs to be consolidated into whole company financial), and so on. Data needs to go only one way from the remote office to the central office. The two servers are on a same logical network, via VPN equipment.
I'm thinking transactional replication with Server A as the central subscriber and Server B as the publisher. The data will need to be partitioned for the most part.
My question is where to place the distributor. The remote office is quite new and still trying to get their systems and operations working. Everyone there is trying to learn how to operate. Because of this, their data may not always be correct and changes (design or just data) are constant.
I was thinking about a separate SQL 2000 remote distributor (relative to the publisher) at the central office. But, is this rather odd, not typically used? Will it create too much network traffic? The reason I'm thinking this type of setting is because having the distributor on the publisher side, I will have no control over what gets replicated over to the main database, right? If something's wrong with the data or with the way it's replicating, I can't really monitor in the middle and stop data from ending up in the main database? The system at the main office is pretty much stable and a lot of its data are critical in nature. Any data discrepancy or problems with system are taken very seriously.
Another question is, with transactional replication ... initial snapshot is typically done. But, since the data to be replicated is only a subset and shouldn't overwrite anything existing before, there should be no initial snapshot? How do I go about getting it to work this way?
Thank you.
May 9, 2008 at 12:43 pm
I'm sure there are better ways, but my got reaction is to create a new SS2k5 database at your central location to be the subscriber. Then you can locally control the content you 'import' into your corporate database. It means extra steps, but affords you extra controls and protections. It also gives you greater flexibility in coping with the database changes out in the branch office.
My 2 cents.
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply