We've implemented aggressive desktop monitoring and blocked downloads from Oracle to avoid the Java subscription. Where it's needed, an OpenJDK distribution is used.
Where we must still use Oracle database, in some small, bespoke legacy use cases (heavy PL/SQL), we've moved to RDS with license included to avoid the direct relationship with Oracle. I get it, a big RAC customer will have a harder time, but they'll also likely have alternatives (e.g. SAP implementation to HANA).
I know of at least one vendor (Hyland) who's dropping Oracle support and providing a migration path to MS SQL. Shame not a FOSS database, but still a trend away from Oracle.
The team defined requirements, ran an RFP and demo process and did site visits to clients of each company. The SAP reference clients weren't exactly thrilled with SAP, the product was too complex and too expensive, but it was rock solid and SAP was a reliable partner. The Oracle reference clients had the usual complaints about features and flexibility, but their real beefs were that Oracle was a predatory and untrustworthy partner.
Oracle made claims in their RFP response that were proven false in the demos and site visits, confirming the claims from reference clients about the company's ethics. In contrast, SAP's RFP responses were validated by the team's due diligence.
So management decided to go with SAP. In response, a senior Oracle person tracked down all of the company's board members and made outrageous claims of incompetence against the company's executives, and alluded ominously about bad faith and conflicts of interest.
Oracle was completely hostile and off the rails when they figured out they lost the deal. I will never, ever do business with Oracle.
Unfortunately, while the SAP application seemed solid, the organization went with their HANA database which was astronomically expensive, and had a bad habit of returning different and provably incorrect results to the same deterministic SQL query every time it ran, and then the entire database would crash for all users.
E.g. you're an IT admin at Big Co overseeing software contracts. You can often get interesting insights by looking at things like how aggressive their sales reps are with end of quarter discounts (how desperate are they to meet numbers that quarter?). Or if you see a company completely dropping the ball within your org, but on CNBC you constantly hear how great the company is by pundits and analysts -- maybe you know something the pundits don't.
Often times the consensus view of a stock trails reality by a few weeks to a month - there's a lot of non-public but also non-confidential information that isn't readily available to analysts, but exposed to employees of customers/vendors/partners/end-users.
TLDR: when stock picking or day trading, pick companies within the niche of the world where you're a SME.
Is it worth the risk/work to move everything over? For a lot of enterprises, their needs to be a huge cost savings or risk reduction. Risk usually being the most important factor the bigger the company.
My understanding is that they were relatively lucky in that most of the hard parts are in the middleware layer and rarely the DB itself - the bank has been around since the 1800s, so has a huge mishmash of technologies that go from old IBM mainframes up to more modern cloud infra. So they're already kind of used to using middleware logic to stitch together various data sources.
The funny thing is that my contact there said the primary impetus is that they see the writing on the wall for a lot of their "legacy" Sun hardware, and figure if they're going to have to redo a lot of it, they may as well re-architect the rest. There'll still be oracle DBs running in the bank for a looong time, but there'll be less and less of it.
In the larger discussion, I also wonder what their new contract rate is for these solutions. Even if 0% were migrating off, if 0% were migrating on then the net rate would still be decently negative because of natural business/app attrition.
What I do see is orgs choosing other Oracle apps like ERP which sneak the Oracle RDBMS in as part of the bundle.
Anyone using Oracle purely as a database is going to migrate to PostgreSQL eventually, but there are a lot of orgs where the database is just one part of a wider Oracle ecosystem with world-class vendor lock-in features.
I worked at a midsize that was core internet infra, where we had an in house OS and ODM hardware and FOSS DBAs. The one Oracle DB and Oracle HW was slipped in the door through finance for ERP as you say. Although I suspect that would be cloud hosted these days.
Think about how hard it would be for you to switch from iPhone to Android. Now multiply that by 10000. That's how hard it is to switch enterprise software.
only after the move is complete and assuming it's as successful as you think it would be. What usually happens is the migration takes on a life of its own and is a multi-year if not multi-decade project. It sucks up so much money and effort that a business could be using to actually build their business vs migration to a different database. Meanwhile, the account execs of the old system know you're moving off of it so say good bye to any kind of contract discounts or special treatment during emergencies.
There's entire graveyards of failed enterprise system migrations. The most likely outcome is eventually a compromise has to be made and now you have two systems to maintain and license, the legacy one, and the new one. With the promise of eventually getting off the old one but it never happens.
I'm on a project with a client that has 24 ERPs across their enterprise around the globe from acquisitions. Half of them are ERPs that were meant to replace another one but the transition was never completed. A big part of this project is integrating all of their sales pipelines, analytics, and history into, yet another, enterprise system.
- Is it possible for a 3 person team to manage 1000 distinct Kubernetes clusters?
- No way in hell!
- What if we hypothetically pay you $2M salary each?
- Well, let me think about it, we could figure this out...
Even if you do move mountains and make it happen, suddenly any outages after the transition become your fault. “This never happened on the old system.”
It took like 5 or 6 years and that $10M represents the cost of only 10 months of operations on Z.
So in such situation, I'd be tempted to actively oppose this initiative.
There was a recent big company that posted on Twitter about "shutting down our last Oracle server" and that was the last thing in a multi-year process or something like that.
Coordination is sometimes harder than the technology itself.
>Market cap of half a trillion.
>Somehow they're "in trouble".
Mega LMAO.
None of those were in business since 1977 (w/ the exception of Nokia, which I would argue is still a successful company today. I wouldn't put it on that list).
None of those were ever valued (even close to) half a trillion, even adjusting for inflation.
Single-sign on, in-person support, certificated software, offering training courses to onboard people, undeletable logs, help with upgrading major versions..
All from a single vendor so you can pick up the phone, yell "fix it" and go on with your day.
Unless they decide to ~~extort~~audit you.
It’s not even really a “chase,” it’s a question of “if I’m building something new, what am I choosing?”
Eventually that momentum can turn into the old thing being worth actively removing.
The languages that get a lot of airtime on HN like Rust, Go, and OCaml are way down in a tier of languages that get a lot of blog posts but enjoy relatively little traction in reality.
It's very surprising to hear you say this, as it's so contrary to my experience.
From the smallest programs (Computer Language Benchmarks Game) to pretty big programs (web browsers), from low-level programs (OS kernels) to high-level programs (GUI Applications), from short-lived programs (command-line utilities) to long-lived programs (database servers), it's hard to think of a single segment where even average Java programs will out-perform average C, C++, or Rust programs.
I hadn't heard of QuestDB before, but it sounds like it's written in zero-GC Java using manual memory management. That's pretty unusual for Java, and would require a team of experts to pull off, I'd think. It also sounds like it drops to C++ and Rust for performance-critical tasks.
Hotspot is the current choice for high performance programs, but is Rust lower performance in some way or are the only downsides related to its younger age?
It’s perhaps useful look at what languages brand new projects are being started with rather than just looking at what languages large established companies like Netflix are choosing.
Companies couldn't care less about the underlying platform or language, they want reliability, stability and tons of easy to find people who can work with it from Day1. Java delivers all that, and will keep delivering for upcoming decades. Big businesses and big money love this (or hate the least out of IT stacks).
A decade ago, a good ~80% of applicants chose to use it or C#.
I personally don't have any issues with working with it, but nobody's learning it outside of work.
On the other hand, it is quite easy to learn, so there's that going for it.