Ok, let's take a hypothetical (though, VERY close to what I've seen). You're a SAAS company. Not a startup, a medium-size company that's been around for 20+ ye

Ok, let's take a hypothetical (though, VERY close to what I've seen). You're a SAAS company. Not a startup, a medium-size company that's been around for 20+ years. Your core software is close to 15 years old. It's 2.5 million lines of code (not that hard to do with "enterprise" software). You have somewhere around 75 TB of data in probably 500 individual customer databases (multi-tenancy was not thought of in 2002 when your core software was developed). This software expects Oracle to be the backend. You can either: 1. Sink time and money into engineering time to evaluate, choose, and implement a new DBMS, and pay for consultants, licenses (if you choose a proprietary one), hardware (you'll need some, because you'll have that weird in-between time as you're migrating your customers' data). You'll have to properly train your DBA's and developers (and ops people) on the new DBMS ($$). You then have to write proper ETL scripts, schedule downtime with your customers (who have nearly 24/7 operations and transactions), make the switch, hope everything went well, then repeat that for your other 499 customers. - or - 2. You can continue to pay Oracle, because you've got that in your budget every year, and it's not worth the hassle.