Connected vs. Disconnected Architecture in ADO.NET
Two ways of talking to a database, and why the right answer depends entirely on what the data has to do.
[ADD — paste the original article text here, or write it fresh in your own words. The published summary covered: what connected and disconnected architectures are, the trade-offs between them, and guidance on choosing one for a given project.]
Connected architecture
[ADD — a direct phone call between application and database: the connection stays open for the duration of data access. Immediate visibility of data changes, low overhead, simple to reason about. The cost is a connection held per concurrent caller, and latency paid per round trip.]
Disconnected architecture
[ADD — fetch, close the connection, work on the data locally, then synchronise. Far better throughput and scalability, and it works offline. The cost is that synchronisation becomes explicit, and consistency stops being automatic.]
Choosing between them
[ADD — the practical rule: connected for real-time, low-volume, connection-cheap workloads; disconnected for high volume, many concurrent users, or anything that must tolerate being offline.]