Should we wait for Oracle 19?

Should we wait for Oracle 19?What a wonderful question:
“We are on Oracle 11.2.0.4 right now. Should we wait for Oracle 19 because of the proposed long term support for this release?”

I have gotten this question roughly 15 times in the past week, raised either by colleagues, customers or partners via email, twitter, in web conferences or (old fashioned way) on the phone. People refer to the graph printed in MOS Note: 742060.1. And I won’t replicate it here as this graph can change anytime. And it got adjusted just a few weeks ago as the support timeline for Oracle 12.2.0.1 was drawn too short.

“We are on Oracle 11.2 – should we wait for Oracle 19?”

Draw a line in the graph in MOS Note: 742060.1 when Free Extended Support for 11.2.0.4 will end: Dec 31, 2018. Just realize please: this is in 13 months.

As Oracle 19 may be available sometime in 2019 probably (that’s why we will call it Oracle 19) this simply means: if you intend to stay please plan to spend an extra premium of 20% on top of your support fee in order to purchase extended support for Oracle 11.2.0.4. Otherwise you won’t get security fixes.

I’m silently assuming that you will do some testing. Therefore 13 months till the end of Free Extended Support aren’t such a long and comfortable period. And therefore the conclusion and answer to this question is:

No, don’t stay on Oracle 11.2.0.4. Move forward to Oracle 12.2.0.1. Start your testing now.

Even worse when you are still on Oracle 11.2.0.3. I’ve just had a discussion with a customer still having many 11.2.0.3 databases. Please remember, 11.2.0.3 went out of any bug fixing support on Aug 27, 2015, i.e. NO SECURITY FIXES for more than 2 years.

Why not wait for Oracle 18 you may ask now. It would be an option as well of course. But there’s no release date known for the on-premises version. If you consider moving to the Oracle Cloud it may be an alternative for you. But there’s no published release date yet. Therefore, for on-premises installations, you clearly should move all your Oracle 11.2 databases as soon as possible to Oracle 12.2.0.1.

“But we are on Oracle 12.1.0.2 already – can’t we wait for Oracle 19?”

Oracle 12.1.0.2 will go out of Premier Support by end of July 2018. But the first year of Extended Support will be waived. So you are safe until mid 2019. Now, before you gamble, keep in mind that you usually won’t jump to the new release on the first day of its availability. You (hopefully) will test a bit (at least).

Look at your current release timelines, i.e. how long does it take you to roll out a new database release. When you are pretty fast you may wait for Oracle 19. Just keep also in mind that often people wait at least for the first Release Update (RU).

You could of course say, we will wait for Oracle 18 and one or two RUs – and then we move. Still fully under Free Extended Support. This doesn’t sound like a bad strategy to me.

Or you could just move to Oracle 12.2.0.1 now. Many of my customers have done this already and surprisingly saw very small number of issues. And then you may skip Oracle 18 and jump to Oracle 19 later on.

Famous Last Words

Please don’t stay on Oracle 11.2.0.4 until the end of Free Extended Support happens. 13 months sound comfortable now. But time flies. You should move forward now. Especially those who have still Oracle 11.2.0.3 databases (or even older releases) in use should move now. Oracle 12.2.0.1 should be your choice.

–Mike

 

6 thoughts on “Should we wait for Oracle 19?

  1. Mike,
    Sometimes, I have serious doubts about Oracle’s corporate leadership. Do they have any idea how much it costs to upgrade an enterprise level database? We are not talking about upgrading a home PC from Windows 7 to Windows 10. The amount of planning and testing it takes is huge. At my workplace, it costs a million dollars (at least) to upgrade from 11g to 18 or from 18 to 19. The databases are not just sitting idle. The applications are constantly being updated and we have to mesh in database upgrades with application releases. The entire cycle takes about one year. Upgrade also incurs database downtime. So, if we keep taking databases down for quarterly patching, Exadata image upgrades, Oracle version upgrades, what will be the up time? To top it all, Oracle does not even provide any release timing for major versions. This creates big problems for planning and budgeting. Oracle needs to come up with at least 5 years of support for any release and focus on providing strong technical support and releasing less buggy code.
    Thanks,
    Arun

    • Arun,

      I see all your points. And we are working on something which may ease your life a lot in the future.
      The benefit of more frequent releases is that it will mean less change as well. Please drop me an email if you’d like to discuss this further.

      My experience after working for 15 years with customers on their upgrades is that most systems can be moved forward quite quickly and with very low effort if a constant process is in place. Of course there are these super huge and important systems which take much more effort. I know that and I see the struggle.

      I think most people won’t take every release step for different reasons. That was the intention of my blog post. But people who still stay on Oracle 11g should think about moving forward, especially with 11.2.0.3 databases for many obvious reasons.

      Again, I’ll happy dig into this conversation with you if you are interested 🙂

      Cheers,
      Mike

  2. Mike,
    Thanks for the reply. The first question of your blog post says it all:

    “We are on Oracle 11.2.0.4 right now. Should we wait for Oracle 19 because of the proposed long term support for this release?”

    Clear indication of what the customers want. They want a release which is stable, supported and they are not forced into upgrading every year. Oracle is going in the complete opposite direction.

    Oracle doesn’t document changes properly. Why would anyone want to read the entire Administrator’s Guide just to see what has changed? If the changes are clearly marked as “New to release 18”, it would make life so much easier.

    Then every change, however small, brings about undocumented side effects and bugs. In the last 18 years that I have been working on Oracle database, there has never been a single new feature which is bug free. Customers do not like to deal with bugs.

    As for the upgrade going faster if a process is in place, it doesn’t work. Oracle keeps changing the way even quarterly patches are applied. So, prior processes have to be re-evaluated. The procedure I used to upgrade from 10g to 11g is not valid anymore for upgrading from 11g to 12c.

    Oracle has really created a bad situation with this yearly release schedule which no one knows when they are going to be actually released and shortening the supported life of each release. Then, in Larry Ellison’s keynote, there was reference to a Data Warehouse version, without any clarification. No one from Oracle even knows what is coming down the line. Keeping the customers in suspense, not very good.

    Thanks again for reading these concerns patiently.
    Arun

    • Arun,

      I don’t agree with your statement “n the last 18 years that I have been working on Oracle database, there has never been a single new feature which is bug free”. catctl.pl has so far (I knock on wood) not seen a single bug filed against the tool itself. And it gets used for thousands of databases already. And we tested it VERY carefully and had a field live production test with the pre version for 11.2.0.3 customers.

      But I fully agree when you say “Customers do not like to deal with bugs.”. I don’t like bugs either. Believe me.
      And I fully understand the frustration involved. Oracle is a very complex product. The complexity has good and dangerous sides. For the good, it is extremely powerful, it offers many options for tweaking and adjusting to deliver the best database performance. The bad is, that complex products and features tend to have bugs. I think this is true for most complex software products.

      But things are always BAD 😉 (Hey, I’m the German here – I’m always negative the stereotypes say 😉 )
      Take the optimizer for instance. In 12.1 we delivered a good product with some not-really-good side effects. You know which features I’m talking about. In 12.2 this seems to work very well now. I’m not saying the Optimizer is perfect but based on the experience I’ve had now with many 12.2. customers I’d say it’s a very cool product – and much better than in 12.1. I can assume, you’ll say now “Why did you deliver the features then?” Simply because the people responsible for it tested it but didn’t take some cases into consideration. Bad idea. I hope we learned from this – and in 12.2 we delivered a much better product.

      I disagree with another statement:
      Not all customers want those super-long support releases. It depends. There are software vendors who say: you are too slow. You don’t offer this and that and I need to roll out a new version on this new OS so why isn’t Oracle … Of course, there are also customers saying “I’ve tested now 3 years with 11.2.0.4 after evaluation options to upgrade for 5 years – and now I will go live in 2018. Sure, you can do this. But the cost are very high as well.

      Oracle 19 will be (that’s the plan) a long-term support release with the option to add Extended Support. That’s a good message. And I can’t tell you how it will be afterwards in the future.

      But I know one thing (and this is important):
      A yearly release means also less development projects. And it means also less change. Which will lead to more stability and shorter testing cycles on customer’s side. At least my feeling tells me that this is a good thing.

      Documenting changes:
      Yes, I know what you mean. I struggle with this as well, especially (and this doesn’t apply to Oracle only) if “security by obscurity” kicks in. Believe me, we do our best. But there are many cases where people (and I’m not speaking about Oracle but generally) develop something and document it later on. Sometimes. Somewhere. I struggle with this every day. In every 3rd project I’m involved in. It’s a personal thing and habit. And this leads unfortunately also to situations where you or I learn about a change in the new release you and I weren’t aware of. We try our best to document things in the NEW FEATURES GUIDE and changes in the UPGRADE GUIDE. But there’s always a chance that something has slipped through the cracks. This happens – it’s not good. But it happens. And we do our best to catch such situations upfront. Just for our own benefit as we are the ones who usually see it first in the first early adopter upgrade projects.

      And finally, when you are on 11.2.0.4, I tried to line out the options. 11.2.0.4 offers the longest support period we’ve ever had for any release. Now you complain that it’s still too short. I tried to explain that you can jump to Oracle 19. But you may reserve some extra budget on the side as you will most likely have a period where no security fixes get deployed. And as I’d assume you’ll wait for one or two RUs you won’t jump when the release gets released. My recommendation would be Oracle 12.2 – give it a try on some test/dev systems. And watch out for something we may deliver sometime next year to help you upgrading faster.

      Cheers,
      Mike

Leave a Reply

Your email address will not be published. Required fields are marked *