Been there. We had a central DB that everyone relied on that also would allow ad-hoc queries from support staff. Locks were an issue and failed batch job would

Been there. We had a central DB that everyone relied on that also would allow ad-hoc queries from support staff. Locks were an issue and failed batch job would leave transaction open. The entire org had to use this database, no way around it and there was 10+ years of legacy code in various systems. We took the simple approach, we had our own database server, so we had the DBAs setup a one-way replication of key tables we needed. We put stored procs and synonyms in front of those tables so in case they moved it wasn't a major expense to us, that also helped the DBAs feel better in case they needed to drop the replication. We also copied over any stored proc we needed to use, pointed them at our synonyms. This also helped to buffer any unintended consequences from an update to a stored proc. Average replication latency was 250ms, setup an alert to let us know if it went higher to let us tell the other teams they might start having issues. The result, the entire company could be frozen from an open transaction on a key table, but our app could keep working. Effort level was low, results were amazing.