I am working with a db and custom application I wrote that is used by many users, all of whom VPN into a server to work. Our team has tossed around the idea of using replication so the users would not have to be hooked up to the VPN to work. I guess we would be talking about using merging, at least I think that is the term.
The db basically keeps track of case manager hours, activities and rates. Users enter their activities, and at certain intervals are prompted to create invoices (over 30 days or $500). When a case manager creates an invoice based on activity records, a series of stored procedures run that figure out the rates and then write some of the activity info to some invoice tables. The invoice is then closed for further activity, processed and exported to the accounting software (user VB functions).
I guess my question is - how and/or DOES replication work when the application interacts with the state of the records in the db to determine where in the process and what next. If, as in my example, the invoice has already been created (by one subscriber) and therefore the items on that invoice are 'closed for editing', what if changes to an item are made by another subscriber (say when they both got the snapshot the items were all 'open' for editing)?
When it all synchs up will it reject the changes made by the second subscriber? If that is the case, subscriber #2 wasted a lot of time making updates that were really not possible. Or worse yet, will the agent try to re-run the sp's to get the records in the invoicing tables match up to the activites. That would make the data in the database different from the printed invoice.
This question is just theorizing, and it is one of the reasons we have stayed away from replication. Any feedback about what I do and don't understand about all of this would be welcome. Thanks!
The db basically keeps track of case manager hours, activities and rates. Users enter their activities, and at certain intervals are prompted to create invoices (over 30 days or $500). When a case manager creates an invoice based on activity records, a series of stored procedures run that figure out the rates and then write some of the activity info to some invoice tables. The invoice is then closed for further activity, processed and exported to the accounting software (user VB functions).
I guess my question is - how and/or DOES replication work when the application interacts with the state of the records in the db to determine where in the process and what next. If, as in my example, the invoice has already been created (by one subscriber) and therefore the items on that invoice are 'closed for editing', what if changes to an item are made by another subscriber (say when they both got the snapshot the items were all 'open' for editing)?
When it all synchs up will it reject the changes made by the second subscriber? If that is the case, subscriber #2 wasted a lot of time making updates that were really not possible. Or worse yet, will the agent try to re-run the sp's to get the records in the invoicing tables match up to the activites. That would make the data in the database different from the printed invoice.
This question is just theorizing, and it is one of the reasons we have stayed away from replication. Any feedback about what I do and don't understand about all of this would be welcome. Thanks!
