Replication (1 Viewer)

[tucker61]

I have a database with 10000 records in it.
It is currently split into front and back end.

Due to network issues I am looking into replication

Questions are
1 do I keep as a split database
2 I have a version control which looks at a table in the back end. How would this work in a replica?
3 trying to add replication I'd field as primary key. Can't do this because I already have data in database. How do I do this?

Thanks in advance.

 

[Rabbie]

1. In a split database you only need to replicate the BE. Each user should have their own copy of the FE.

2. This should work OK provided you do a replication/synchronisation when you update the version.

3. Not clear exactly what you mean here

 

[tucker61]

I was thinking of replication because the data page on my front end is daily showing the #deleted comment in my fields.

So I think the network may be corrupting the data. (the database is probably the least priority database on the system, so can't get any system support)

I could replicate the back end only and leave the back ends on the local servers, but since the people who would be using the database are over
200 miles apart, across 5 sites, and they cannot see each others local sever, then this would be a mammoth task to maintain.

The other thing I do not like about replication is that you the have to rely on the individuals to sync back to the master. I am not sure if this is possible to do via code?

Point 3 is trying to get autonumber to work when I already have data in the database.

 

[Rabbie]

See this thread for more help on replication

 

[dfenton]

Replication is never a solution for performance or network unreliability problems. All you would accomplish in your case, where it appears to be an unreliable network, is to move the problem to a place where it is substantially more dangerous. That is, you still have to synchronize the local data, and if the network problem pops up during the synch, you can corrupt the remote database. The way to avoid that risk is to go with indirect replication, but that's overcomplicating things since it requires setup on all the client PCs and on the server, and needs to be carefully maintained.

Fix the network problem and forget about Jet replication.

On the ID issue, GUIDs are not good for PKs (not sure if that's what you were using or not), for the reasons outlined here:

http://trigeminal.com/usenet/usenet011.asp?1033

On the issue of randomized Autonumbers, if you care about the value in an Autonumber field, then you're misusing Autonumbers.

If you need to maintain a sequence, then the problem is fairly complex. I won't go into the explanation unless that is your actual question, but in general, I'd say you should NOT use Jet replication in your scenario at all, but instead fix the bad network.

 

[Simon_MT]

A terminal Server is another consideration and providing images are not used it is efficient. All it is does is sending screen dumps accross the WAN. You only need to put a copy of the FE for each user. You can also integate all documents and email at the same time whilst allowing their local printer to be set-up on the start of the RDP session.

Simon

 

[dfenton]

Windows Terminal Server/Citrix and the RDP client are much more efficient than sending screen dumps (i.e., bitmaps) -- they use the Windows low-level graphics primitivies, i.e., the instructions Windows sends to the graphics driver, which are then processed and translated into actual pixels for the monitor. In an RDP session, the remote server sends those same low-evel graphics instructions, and then your local PC's graphics driver translates it into the exact pixels needed to paint the screen.