pricerc wrote:if there would be a case for a 'client only' installation for Jiwa that excludes all the SQL 'server' components? Obviously, this version would be unable to create databases (even a demo one), but that may have uses outside of this discussion: e.g. in a larger network, it may be desirable to have a client install rolled out in a group policy, and they may not want people to be able to create their own databases.
You're making my palms sweaty just thinking about what this would involve! - not just the application, but increased burden on support, documentation and the like in dealing with the introduced complexities. The idea has merit - but the business case would need to see fair bit of demand for this to be worthwhile. As much as I love a good challenge - and this would be a fun project - I'm sure our bean counters would frown upon putting resources put into that without solid evidence of demand.
We are looking to making Jiwa more multi-tenant friendly for RDS environments - the main downside there being a need for an internet connection for customers to access their Jiwa. But if that's not an issue, I envisage a scenario where one RDS machine could serve multiple tenants (Jiwa customers) in a partitioned, secure and seamless fashion.
Obviously, you'd need to have a number of users across all tenants before you reach the cost effective break-even point - depending on the hosting costs.
I wouldn't consider multi-tenanting the databases - they'd need their own. We certainly could achieve that with introducing a tenantId per table and implementing row level security - but that's a fair amount of work and a lot of interesting problems would need to be solved, so I'm not entertaining that aspect of multi-tenanting.