Yesterday I put out a post called Does anyone care about enterprise application servers anymore?
Before I posted it I got my boss to read it. This is how the Teams chat went.
- Boss: It seems fine. Was there anything in particular you were not happy with?
- Me: I guess I’m a little nervous about the post because the obvious comeback is, “Isn’t that true for the Oracle database too?”, and to some extent I would have to say yes*… I did consider adding a paragraph about that, but thought I would rather not.
- Boss: Actually that is what I was thinking while reading it.
* In the context of the chat, “yes”, meant is was a valid question. Just not easy to answer.
Sure enough, not long after posting it I got a Twitter response from Timur Akhmadeev that said, “If you replace ‘enterprise application servers’ with ‘Oracle DB’ – does your message stay same? or different?” That is the title of this post. 🙂
I’m going to attempt to explain my thoughts on this by going through the same talking points.
I’m an Oracle fanboy. You know that. My take on this subject is bound to be biased because of my history.
I know I am mostly known for Oracle stuff, but in my current job I have to look after MySQL and SQL Server databases. I work on one project that uses PostgreSQL, which I’m really bad at. The company recently started using Snowflake, and the plan is to move all analytics and warehouse type stuff to that. So as much as I’m an Oracle fan, my work life is a lot more diverse than you might think.
Third-party application vendors
One of the major points in my original post was about the vendors of third-party products that got rid of enterprise application servers from their stack and replaced them with more streamlined alternatives. Do I see this reflected in the database space?
From my perspective no. We have some products that are 100% Oracle. Nothing we can do about that other than move to completely different products. Of those where the RDBMS engine is a choice, I don’t think we have one where they offer other RDBMS engines, but not Oracle. It seems Oracle is still a viable choice here. I’m not saying they are recommending it, but equally they are not advising against it.
There are some vendors where MySQL is not an option, but PostgreSQL is. Interesting…
Overall, we don’t see the same pressure from the vendors in the database space as we do in the enterprise application server space.
Similar to my comments about application servers, most RDBMS engines can work within containers, but should they? It’s fine for a demo or a small system. Are you happy about running a huge data warehouse as a monolith in a container though? I don’t think this is about Oracle or not. It’s something more fundamental than that.
There is definitely a big learning curve for Oracle. I could argue it’s probably harder to get good at Oracle administration than many other engines. I think you have to be really careful of the comparison though. Take my current company for example. The big and complicated stuff happens on Oracle, and the really simple stuff happens on SQL Server or MySQL. It’s unfair to compare the administration requirements because the use cases are so different.
I think the only way you can compare them is if someone has legitimately done the same types of projects on multiple engines and really understands them. I’m not that person, so I can’t make that comparison.
But what about product X?
As with the application server comment, if you have a vendor/product that demands a specific engine, you have no choice but to stick to it, or change vendor/product entirely.
For new products I would suggest you get some information about their current customer base and go with the majority/trend. If Oracle is an option, but 90% of their customers are not using Oracle, I would not knowingly make myself a minority without good reason.
Moving to the cloud
The cloud is the great leveller, and I think this is both good and bad for Oracle.
In the case of administration, cloud services can hide a lot of the complexity and make things much more generic. Take a look at RDS on AWS and you will see MySQL, SQL Server, PostgreSQL and Oracle all feel quite similar from a basic administration perspective. If people think Oracle is too complicated, take a look at RDS on AWS, or Autonomous Database on Oracle Cloud.
Where it becomes an issue for Oracle is as soon as the guts are hidden from us, how valuable is the Oracle IP? Imagine feature X is baked into the Oracle database, but requires 3 additional apps to achieve the same result with another RDBMS. On-prem this is a real pain, as it’s now my responsibility to get all this stuff working and keep it working. If a cloud service does all that for me, do I care if it is baked into the RDBMS or not? I don’t think I do. I just care about what features the service delivers and the price.
There is also the comparison of “built for the cloud” vs “monolith in the cloud”. I know this concerns a lot of people, but to be honest I don’t care. If the resulting service gives me what I need, how much do I care what’s under the hood? That’s the cloud provider’s problem, not mine.
Regardless of which engine you pick, I think you should assume you will be using a cloud database service unless there is a compelling reason not to, rather than the other way round.
RDBMS vs NoSQL?
I’m not going to get into the RDBMS vs NoSQL war because I think time has proved it to be redundant. I’ve been listening to this for about 15 years and it seems pretty clear to me the choice depends on the use case.
Some use cases fit really well with NoSQL and some don’t. There are also grey areas between where polyglot engines become interesting, but it all depends where your use case fits on the spectrum…
Imagine a shop that is mostly Oracle, but needs a JSON document store. Do they use a different engine like MongoDB, or just continue using Oracle and use SODA? The answer will be different depending on how you frame the question.
What’s the real problem with Oracle?
I don’t think Oracle’s problem is their tech, or even their prices. I speak to a lot of people and the number one problem they have with Oracle is the shady business practices. People don’t trust Oracle. That is not a good thing to hear from the customer base, especially when there are more alternatives than ever.
Oracle is in dire need of a Microsoft-style transformation, and that can only happen if the senior management at Oracle really want it to happen.
Is there a conclusion?
When you consider the context of the original blog post, not just the title, I don’t think you can just substitute “Oracle DB” for “enterprise application server” into the question. It’s too simplistic.
PS. The answer is, “it depends!” 🙂