Oracle Autonomous Database and the Death of the DBA

Before we start, make sure you watch the keynote and actually listen to what is said from a tech perspective (not the pricing stuff). Don’t prepare your counter argument before you hear what is said. Listen!

Myself and many others have been talking about this for over a decade. Most recently in my post before this announcement called No DBA Required.

As a cloud provider, Oracle want to make products as autonomous as possible because it reduces the amount of bodies they need to look after them and reduces human error. Yes, even the best of us screw up! It’s also annoying when you have a good product and stupid people use it badly, then blame the product. The aim of this suite of autonomous databases is to automate as much as possible to reduce the need for human interaction and free us from the mundane. There is an important slide at about 41 minutes that sums this up well. Less time on infrastructure, patching, upgrades, ensuring availability, tuning. More time on database design, data analytics, data policies, securing data. With the possible exception of tuning that can be interesting, let me put this another way.

Less time on boring shit. More time on important shit!

This is firmly targeted at removing the need for operational DBAs. A role that *you* should have already automated out of your organisation anyway. If you’ve not, then you have failed.

It should be obvious that the Development DBAs are sitting pretty here. The thinkers are safe. The recipe followers are not. Your mantra from now on should be…

Keep learning. Keep improving yourself. Keep your job!

For context it’s worth mentioning a few things.

  • Oracle 18c is effectively 12.2.0.2, but as has happened with recent patchsets, it contains a bunch of new functionality.
  • Although Oracle 18c has new features that make it easier to build an autonomous database, the “Autonomous Database” is a cloud service, so you will need to run your database on Oracle Public Cloud, or possibly on Cloud@Customer. Just installing 18c on-prem will not get you an autonomous database.
  • This is the first generation of the first service in a suite of database services for a variety of workloads. The more customers that use these services, the more workloads Oracle will see and the more autonomous they can make them. This is the beginning of a process, not the conclusion.

My view? Bring it on. Save me from the boring crap! 🙂

Cheers

Tim…

Update: Rather disappointingly, but as I expected, many people don’t seem to be able to look past the name of the service and are just making knee jerk reactions.

Also, there seems to be some perception that I’ve drunk the Kool Aid on this. I’ve been looking at the information about the service and I think it is interesting. Over time I will refine my opinion, as the facts emerge. What I won’t do is write it off before it’s even been released. It’s that sort of attitude that has turned some DBAs into inflexible dinosaurs. You’ve got to evolve or die people!

Update 2 : I’ve now tried the service. You can read what I thought here.

Author: Tim...

DBA, Developer, Author, Trainer.

28 thoughts on “Oracle Autonomous Database and the Death of the DBA”

  1. I am crazy upset! I started to write almost the same thing and you beat me to it. The death of the database babysitter, the “keep up with technology -who me?” DBA and the “Make my resume look good and BS my way through the interview” DBA(Database Apllicant – not to be confused with Admisistrator) have been coming for a while now.

    I argue that complexity is (and will be) the driver – automation is only one bullet point in the list of topics under the umbrella of complexity. While automation seeks to reduce complexity in one sense, I think we may find that it just pushes those costs elsewhere. This will result in a requirement for resources with higher technical depth in the long run. THAT is good news for the Tim Hall’s of the world!

    #OOW2017

  2. My old company always had poor operational DBAs (due to management really) and did everything to make development DBAs disappear except in a few holdout departments. Ellison’s current message is very welcome in that regard – much better than his previous “without changing a single line of code” dig that targeted development DBAs and developers. Yes, Maria, I remember InMemory…

    That being said, he purely and simply contradicted himself:
    – on the one hand, he said DBAs would always have enough to do but only the interesting stuff would remain.
    – on the other hand, he promised huge reductions in labor costs!

    How can you reduce labor costs without letting people go?

    Ellison is in competition with *everybody*, including his customers, the DBAs who make his products work and the developers who make them worth the money they cost. He can’t help it.

    See the bright side of the new situation and the new opportunities as much as you want or can, with my blessing. But admit that Ellison is trying to pull a fast one.

    Best regards, Stew

  3. “Just installing 18c on-prem will not get you an autonomous database”?
    You mean unless we outsource our production workload to Larry’s “cloud”, we won’t have an “autonomous database”?
    Ah, OK: back to square one of good old release 8.
    Thanks but, no thanks.

  4. Well. What if automated patch update will crash DB or will lead to application misbehaving? Who will be responsible on this and pay for business losses? And who will restore everything back?

  5. Noons: It will not be right for every customer. Choice is a wonderful thing. 🙂

    Nikolai: From and operations perspective it will be a black box. If you sign up to this service to are handing control of this over to Oracle, with all the benefits and risks associated with this.

    Cheers

    Tim…

  6. Nice post. It will bring new opportunities for great DBA’s & they will transform themselves. DBA’s who will deny to reskill might die..

  7. Wonderful post, and I think you know where I stand on this topic. There will always be a need for data professionals. There is going to be less need for someone to handle backups, restores, and even performance tuning. These days the only people I see clamoring about how there will always be a need for a traditional operational DBA are the people that make a living by charging by the hour to be that DBA.

  8. At last I was able to see Larry’s presentation.

    Very interesting video, but all these talks about security, especially about detecting correct and incorrect SQL-requests, make me laugh – if a 3rd-party system has ability to read and scan queries and DML it means that this particular system can restore your data from scratch and you won’t ever know about it. The best security solution you can ever imagine!

    And, by the way, Larry said (very shortly), that this security system will detect an attack and patch database. But most attacks need to be immediately reacted to and be blocked before DB patch. Any ideas on this?

    Continuous performance self-tuning. Also sounds very impressive. But I suppose every DBA in his life meets such situation, when Oracle advisers and ADDM make such terrible advises, that might lead even to unexpected downtime. In my last scenario they advised to make a bitmap index over frequently updated table.

    I’ll stop here, probably later I’ll make a post with my vision – this topic is very wide.
    But I disagree – DBA’s won’t die at least for the next 10-15 years.

  9. Nikolai : I don’t disagree with you, but I think you are locked into a frame of mind that means when you hear “Operational DBA” you assume it means all DBAs. It doesn’t. The slide at 41 minutes clearly showed Oracle see a role for DBAs in the future. It’s just not doing the mundane crap.

    As for the performance, when Oracle start showing the numbers you will see they have tested this with real customer workloads and it has worked well. Is it perfect? No. Will it ever be perfect? Probably not. Will it free you up from doing boring crap? Yes, I think so.

    I see the end of what we traditionally call DBAs, because that includes so many tasks that are no longer necessary these days if you are doing your setup correctly anyway. As Thomas (SQLRockStart) above says, there will always be a need for Data Professionals. I would hazard a guess that you are already a data professional, rather than a DBA in the style of the 1990’s. 🙂

    Cheers

    Tim…

  10. Here’s the thing. Automation is great for handling the routine, mundane tasks in database administration. It can even take over in addressing common errors. But, at least for now, you still need DBAs to handle the exceptions.

    For now and based on what we currently know, this will allow more scaling of administrators – fewer DBAs can manage more database instances. But, as you point out, we’re still in the very early stages of these developments. Stay tuned.

  11. Great post Tim. I like how you take the positive approach and highlight the need for people to keep learning and improving their skills. The less time we spend on the “boring stuff” as you mentioned, the more time we spend on the “important stuff”! Which is the entire point of software 🙂

  12. Now DBA ,Need to get new skill set because it is good to mange large scale dbs by oracle engineering team. Defiantly a dba will play his role.No need to be worry

  13. That was what I expected before , The change is a big part of our career life,
    I was really eager to learn much about oracle database even I had grate experience with oracle apex.
    but every thing is changed .
    I think it is time to abandon Oracle Database Administration job , I believe if someone want to understand Data science , He should be first of all a Grate DBA , So we have good Background to continue working in Data Field . we should not limit ourselves to work on oracle technologies .
    this is a grate opportunity , no need to be worry.

  14. Thanks Tim!
    Just wondering what of the database tuning scope will be integrated into the “automated tuning” in 18c. SQL query tuning, oracle CBO setting tuning, init parameter tuning?

  15. Daniel: 18c is not the thing that is autonomous. It is the cloud service built over it that is.

    18c is basically 12.2.0.2, so it is bug fixes and some nice new features. It is not a revolution. It is the cloud service that is the big deal.

    Cheers

    Tim…

  16. Being a DBA I am ready to welcome new changes in Oracle world that will be most beneficial for us. Thank you, Tim, The above information is useful for us.

  17. Great post Tim and others useful comments.
    I want to know what really now we Dba’s have to focus on? Please be in details.
    Secondly I have pretty good knowledge in SQL and Pl/SQL, can it help in development or it is also going to be automated?
    May be these are not good questions but let me clear my doubts.

    Cheers!!

  18. Piyush: It’s impossible for me to say what you should focus on. It depends a lot on your interests and background. I could tell you “Data Science”, but unless you have a background in maths (Statistics and Probability), you probably won’t be able to do it. I could tell you to look at development, but if you don’t enjoy it you probably won’t work hard enough at it to get good. You have to find something you enjoy and commit time to it.

    Regarding SQL and PL/SQL: I discussed this here.

    https://oracle-base.com/blog/2016/08/13/real-talk-plsql-and-sql-as-your-only-development-skill/

    Cheers

    Tim…

  19. Great Post by Tim and useful comments from all others.
    As a oracle DBA if we can able to learn new skill set like Exadata and Hadoop is it better for us to sustain in IT Industry now

  20. What is the scope for hadoop in near future? Will automation also impact hadoop professionals job in future like oracle DBA’s now? Please let me know all of your views

  21. Shrinath: If something can be automated, it will be. If you’re planning on making a career on managing infrastructure for anything, I would probably think again. The people that will last the longest are those that can add value beyond what a script can do. 🙂

  22. Thanks Tim for your valuable reply. 🙂

    So, what according to you would be better profession ( among hadoop(any specialization), devOps, AWS, etc) to survive in IT world for next 15 to 20 years? Please elaborate your views with some reasons (if possible) as I am bit confused on what to do now? Where should I head on? Just FYI, I was good in coding (logic development level) and mathematics during my college days.

  23. Shrinath: I have no answer for that. Do what you enjoy and keep an eye on your section of the industry to see how it is changing over time. There is no reason to panic. Just evolve.

    If I knew and answer, I would also be making that move. 🙂

    Cheers

    Tim…

Comments are closed.