4 ms·
> and rewrote all their products to be SaaS. I had a PoC with DBA's from my company and Oracle to see how that works for simple Oracle DB use cases and some E
by szszrk 6y ago
> and rewrote all their products to be SaaS.
I had a PoC with DBA's from my company and Oracle to see how that works for simple Oracle DB use cases and some Exadata stuff. At the time (2-3 years ago) most of that "rewrite" was just a lot of scripts on their side that reconfigured regular Oracle DB. Same work my DBA's did. They even allowed us to modify managed DB's configuration, but apparently it meant we'll loose support for that and would have to start managing it by ourselves again. It was confusing at best and their team was sometimes as surprised as we were.
Still, it's better then "classic" they offered. Smallest Weblogic setup (we wanted two servers) that was possible to create using a wizard was around 70GB of RAM (AFAIR) and multiple servers. It just spawned full VM with managed Oracle DB for Weblogic configuration, another one for DR, two WLS servers, a loadbalancer. Take into account that the smallest VM was 4 or 8GB of ram and your budget is gone right away. Not really for small scale setups.
I wonder how it looks like from perspective of people more fluent in Oracle products and OCI cloud, currently.
- toyg 6y agoI’m not familiar with every product, but I worked there and I can guarantee you that a lot of their products had to be substantially rewritten. The scenario is somewhat confused by the fact that they try to sell everything as “cloud” but some are really just old-fashioned hosting with a bunch of scripts as you said. In the worst cases it wasn’t even scripts, it was someone in Asia getting a notification to do some work... The final objective is “real” SaaS / IaaS though. Making stuff cloud-first for years was the only thing that mattered internally, a massive undertaking that took ages and might not even be done yet for some products. It burned their partner ecosystem almost to the ground, anybody who was in infrastructure basically had to gtfo or be a dumb reseller; and a lot of on-prem customers moved to other products (because Oracle basically stopped updating on-prem, or dropped crippled bundle-patches years after deploying new features on cloud, sending a clear message that on-prem was dead. Faced with a migration to what were often significantly-different cloud versions, a lot of customers opted for a migration to better products from the competition...).