I have a weird scenario and I can't seem to find the best way to make it work.
I have a inventory app stored on an Linode server. This app handles different companies. Each company has its own database.
All companies have multiple stores located at different locations. All stores need to use the same app and at the same time the data has to be synced.
I need to replicate the data, But all stores/apps need to be able to write/read and replicate at the same time. The problem is that most of them don't have Internet connection for hours. They are totally disconnected from the world(just LAN).
The conventional MySQL replication is not going to work because it needs internet connectivity to stay operational.
What do I do???
Is having my own software solution that replicates data on a higher level a good idea? If yes are there any best practises I should follow?
I also can't use mysql auto_increment step and offset for ID generation because some of the clients keep opening more and more stores. Do I need to generate my own GUID for each entity to make sure ids don't clash by prefixing the store unique ID (STOREID-UNIQUEID)?