Oakies Blog Aggregator

Autonomous Database

Larry Ellison has announced Oracle 18c to be the Autonomous Database, the no-human labor, and self-driven, database. Here is a quick recap of what it is behind the marketing words. My opinion only.


Since Oracle decided to be a public cloud PaaS provider, they announced ‘unmanaged’ and ‘managed’ services. The managed service for DBaaS is where you are not the Database Administrator. Autonomous Database 017-10-04 15.51.53 You have full administration right to deploy and develop your application, but the system administration is done by Oracle: provisioning, patching, upgrade, system tuning, availability. The first managed service was announced last year at OOW16: the Oracle Exadata Express Cloud Service, for developers. And the second managed service is the Oracle Data Warehouse Cloud Service, for data warehousing. It is planned for December 2017 and will be based on new version: Oracle 18c.
And let’s be clear, the ‘autonomous’ label came at the last minute, as the marketing message for Open World, to show that the managed service is fully automated, because Cloud is about automation.

So, is that only marketing? I don’t think so. There are real features behind it. And some of them exist for a long time, and had just to be enhanced further in 18c.


The features are there for a long time for full automation. RAC protects from instance or server crash and Data Guard protects from all other failures. Both are fully automated with no-human intervention: service relocation in RAC, Fast-Start FailOver in Data Guard. And both can failover transparently with Application Continuity. Oracle also recently introduced Sharding, to link data availability to different datacenter. And this technique will also be used for RAC instance affinity in Oracle 18c.

Patch and Upgrade

You can apply patches in a rolling manner with RAC. The problem was OJVM, but this will be also rolling in 18c. You can do rolling upgrade with Data Guard. And 18c will have automated and faster upgrade at PDB plug-in. From a system point of view, all is fully automated. However, we will still need human intervention for testing, and planning it at the right time, and of course for troubleshooting when something goes wrong. The autonomous database is not incompatible with that. With logical replication (Golden Gate, Transient logical standby) or versioning (Edition Based Redefinition, Workspace Manager) Oracle has the tools to automatically provide the environment to test the upgrade before it is opened in production.

We can also imagine that other features may help to avoid regression. For example, SQL Plan Management can prevent execution plan regressions on common use-cases, and let the (human) user accept evolved plans later. This can also be done automatically (but humans still have to define the capture). Of course, we all know the limits of the advisors and automatic implementation. But there are also some applications where it can be fine. This ‘autonomous’ Cloud Service is a possibility, not a universal solution.


Again, we all know that the database cannot be optimized without the knowledge of the data and the design of the application. But Oracle has also a lot of features to automate some common practices. Automatic Data Optimization, Segment Advisor, SQL Access Advisor, Auto DOP, automatic choice of buffered or direct reads,… We have seen a lot of drawbacks with SQL Plan Directives, but that was mainly in OLTP. This new managed service is for DWH where dynamic sampling is not a bad idea.

The idea is to ‘load data and run’ and Oracle takes care of index, partitioning, etc. You create the table and declare Primary keys and Foreign key (RELY DISABLE NOVALIDATE see a previous blog post about that). Then I suppose that Oracle can guess which are the dimension tables and the fact tables. And then do some common things to do on that: partition on the date dimension (if there is only one – maybe it has to detect some load/query patterns), create bitmap indexes on all fact foreign key. Online statistics gathering will be extended in 18c to incrementally maintain statistics on bulk-insert, and this may include histograms.

I’m very skeptical on that point, because I’ve seen lot of datawarehouse databases where, even on big hardware, there is always a need for optimization. But my point of view may be biased. I’m a consultant, and then I see only the databases where people think they need human analysis and troubleshooting. There may be some non-critical datawarehouse databases where nobody is doing any optimisation, and then implementing some default optimization may be sufficient to make it a bit more efficient. This autonomous elastic cloud service may be a good start for some projects, when it is difficult to plan the hardware and human resources that will be needed. But I’m quite sure that after a while, designing an efficient and scalable infrastructure and data model will still require our DBA skills and human intelligence.


This service is provisioned as a PDB where what we can do is limited by the multitenant lockdown profiles. We can connect easily (with a credentials .zip) from SQL Developer, and we can load data from an object store using the DBMS_CLOUD package to define the credentials (Swift) and load text files. Once again, it is an evolution of existing features like external tables and preprocessors.


This service is Elastic: it can scale the CPU resource up and down without stopping the service. Again this can use existing features: OVM for the host, and Resource Manager for the CDB and PDB.

So what?

Some will say that it is only marketing with nothing behind, and the same announcement that was made by each previous version. Others will say that it is really autonomous, self-driving, self-securing, self-repairing. This ‘self-driven’ idea is an analogy with Tesla (Elon Musk is a big friend of Larry Ellison), but for the moment, there’s still a human inside a Tesla. Autonomous is a vision for the future, not the present. The present is a new managed service, more online operations, and easier upgrades.

What I really like is the integration of existing features to serve a new marketing trend. Do you remember when the flashback features came out? Flashback query existed internally since the invention of rollback segments (and maybe even before with the Before Image). It was exposed in 9i with a dbms package, and 10g in the SQL ‘as of’. All was already there to bring a feature that no other RDBMSs are capable of. The existing features support the marketing message at the right time, and this marketing message encourages to develop new features again, like new online operations. And those do not benefit only to the managed cloud services as they go to the main branch of Oracle Database.

The other thing I like is the idea to have a configuration dedicated to specific needs. In the OOW17 hands-on lab, there was even a “_cloud_service_type” parameter set to DWCS. Here this service is for DWH, and there will be one dedicated to OLTP mid-2018. This is something I would like to see in future versions. For example, there was a lot of discussion about 12.1 enabling adaptive statistics by default, and 12.2 disabling them. This kind of default settings could be set depending on the database type: more conservative for OLTP upgrades, more adaptive for new datawarehouse projects.


Cet article Autonomous Database est apparu en premier sur Blog dbi services.

What I Learned At Oracle Open World 2017

My Oracle Open World started with a beautiful Sunday-morning run across San Francisco’s Golden Gate Bridge annual event organized by Oracle’s SQL Developer team (thanks Jeff)! Monday morning another group met for our annual San Francisco Bay Swim at the Dolphin Swim club; thanks to the Oracle Developer Community for providing coffee, pastries, and cool towels (thanks Jennifer)!

Oracle’s Open World was a little cozier this year; most sessions were at Moscone and hotels in close proximity making it easier to get from event to event. More time in sessions and less time in transit; good swap.

The conference halls were huge and full of software I was familiar with and several things that were new to me; Oracle has over 6,000 products so it was interesting to see what’s new. I love to spend time at Oracle’s kiosks talking to the product teams and seeing live demonstrations all the while asking many questions; thanks to the many people who put up with all of my curiosity and questions.

First up was User Group Sunday where I sat through some excellent sessions sponsored by IOUG (one of them mine), ODTUG, and OAUG. Once again the user groups provided solid technical content with no marketing to be seen; thanks!

Oracle’s Larry Ellison’s first keynote on Sunday talked about Oracle Cloud (surprise!) and the coming of the “Autonomous, self-driving” database; a session provoking much discussion. This was followed up the next day more-completely by Andy Mendelsohn, Oracle Executive VP of Database Server Technologies. One option for Oracle Cloud customers will be to use the Autonomous Data Warehouse Cloud (more coming later) where much of the “grunt work” some DBAs perform is automated freeing people to spend time solving problems and helping their organizations use data. If you are one of those “DBA”s whose day-to-day functions can easily be automated; it’s time to learn some new skills. If you’re one of the more-talented DBAs you’ve probably already automated most of what you do and you’ll welcome the built-in automation (watch for a more-detailed article about this soon).

Of course most of Oracle’s message focused on the Cloud, more Cloud, and yet more Cloud; they’ve done lots to improve and simplify our interactions with the Oracle Cloud making it more useful each year. Some items of great interest are the new Universal Credit and Bring Your License PaaS options.

Another big announcement is a change to Oracle’s release and naming convention for the future. The next release of the Oracle Database (currently 12c R2 or 12.2) will be 18.1! Yes, they are changing the numbering scheme and the way they release products to ensure a more even and more rapid flow. Annual releases (e.g. moving from 18 to 19 will be upgrades), “dot” releases will be patches. We can expect regular quarterly Release Updates (RUs) containing bug and security fixes; Release Update Revisions (RURs) will occur as needed and will occur urgent fixes and regressions fixes. For now all will be handled with the familiar OPatch mechanism. For more information see Release Update and Release Update Revisions for Database Proactive Patch Program (MOS Doc ID 2285040.1) These changes reflect Oracle adopting a continuous delivery model in keeping with the industry DevOps trend.

Speaking of the database, Oracle’s doing more then changing the name; the 18c release will include many new features and improvements; I’ll publish another article next month with many details!

As a developer, one of the coolest videos and keynotes this week focused on the “Bloodhound” 1,000 MPH car and the amazing data supporting it. Oracle has teamed with The Bloodhound Project and its own Oracle Academy to make the data involved in the project available interactively to millions of school kids around the planet. An amazing way to get kids interested in Science, Technology, Engineering, and Mathematics (STEM).

Of course, lots more happened and I’ll elaborate more in future posts.

12c Parse

Following on from a comment to a recent posting of mine about “bad” SQL ending up in the shared pool and the specific detail that too much bad SQL could cause contention problems while staying virtually invisible, there’s a related note today on the ODC (formerly OTN) forum of a little change in 12.2 that alerts you to the problem.

Try executing the following anonymous block (on a non-production system):

        m1 number;
        for i in 1..10000 loop
                execute immediate 'select count(*) frm dual' into m1;
                when others then null;
        end loop;

Then check your alert log (if you want to be a little cautious, change the 10,000 in the loop to something like 200). If you’re running you’ll find something like the following:

ORCL(3):WARNING: too many parse errors, count=100 SQL hash=0x19a22496
ORCL(3):PARSE ERROR: ospid=4577, error=923 for statement:
ORCL(3):select count(*) frm dual
ORCL(3):Additional information: hd=0x7673c258 phd=0x765151a8 flg=0x28 cisid=135 sid=135 ciuid=135 uid=135
ORCL(3):----- PL/SQL Call Stack -----
  object      line  object
  handle    number  name
0x76734f18         5  anonymous block
ORCL(3):WARNING: too many parse errors, count=200 SQL hash=0x19a22496
ORCL(3):PARSE ERROR: ospid=4577, error=923 for statement:
ORCL(3):select count(*) frm dual
ORCL(3):Additional information: hd=0x7673c258 phd=0x765151a8 flg=0x28 cisid=135 sid=135 ciuid=135 uid=135
ORCL(3):----- PL/SQL Call Stack -----
  object      line  object
  handle    number  name
0x76734f18         5  anonymous block

The warning will be repeated every hundred occurrences. As you can see the guilty (ORA-00923: missing FROM) SQL appears in the report so you know what you’re looking for. In my particular case, with the silly PL/SQL block, the address of the calling anonymous pl/sql block was also reported:

select sql_text from V$sql where child_address = '0000000076734F18';

declare m1 number; begin  for i in 1..10000 loop  begin   execute immediate 'sel
ect count(*) frm dual' into m1;   dbms_output.put_line(m1);  exception	 when ot
hers then null;  end;  end loop; end;

In the case of the OP on ODC the SQL reported in the alert log was simply: “SELECT 1”. As Billy Verreynne suggested in the thread, this looks like the sort of code that would be sent to the database by some of the connection pooling clients to check that the database is up. Unfortunately (apart from the waste of effort) this particular setup seems to think it’s talking to some database other Oracle!


This is a feature of 12.2 – 11g and 12.1 don’t write such warnings to the alert log.


A tweet from Mohamed Houri reminds me that parse failures like these, of course, show up in the instance activity stats, in particular:

Name                               Value
----                               -----
opened cursors cumulative         10,006
enqueue requests                  10,002
enqueue releases                  10,002
sql area purged                   10,000
sql area evicted                  10,000
parse count (total)               10,008
parse count (hard)                10,002
parse count (failures)            10,000

The enqueue requests are for the ‘CU’ (cursor) enqueue which, I think, appeared in 10g – they’re acquired (and released) on every hard parse.

Most of the figures that my session reports here are likely to be highly camouflaged by the rest of the activity from a normal system, so the most important number is the “parse count (failures)” – so it’s useful to know that you can subtract that number the other statistics to give you an idea of the impact that would be eliminated if you could located and stop the thing generating the failing statements.


Patrick Joliffe (see pingback below) has published an article pointing out that in earlier versions of Oracle you can set event 10035 to get the same information dumped into the alert log on every parse failure.


Where in the World is Goth Geek Girl, Week 41

So now that I’m finished just taking it easy, playing hooky from Oracle Open World and Oak Table World…uh, yeah….

I get to recover from emergency surgery for another week and while I’ve been busy writing slide content, building out new abstracts for the new year, along with content for a couple books, I’m finishing chapters on.  My ebook is about to come out for DevOps for the DBA: The Next Frontier, (I’ll be giving a few hard copies away at events in the next couple of months, but you can get a free copy at Delphix.com soon!)  This is an enhanced version of my blog post series that’s been running on Delphix’s site, including updated content surrounding DataOps. Along with all this, the awesome Dominic Delmolino and I’ve been trying to finish up a podcast.  I consider Dominic one of the father’s of database DevOps, (the man’s the same age as me, so hopefully he won’t be insulted by the “father” description…:)) I’ve also been busy with the demands of local user group board demands and for my upcoming schedule, there’s a few things in the next week to consider:

October 11th:

I’ll begin with a webinar next week for the SQL PASS Virtualization Group, doing a session on DevOps for the DBA.  This continues the ongoing discussion on the future of the DBA career, which many have been calling, DBA 3.0.

October 14th:

I’ll be presenting at SQL Saturday Charlotte with more DevOps for the DBA, this time in person!  If you’re part of the SQL Server industry, you should be part of SQL Pass, it’s free and there’s so much regional activity.  The education opportunities for SQL Server professionals is really impressive, so make sure to check it out.

October 18th:

I’m doing another webinar, this time for the SQL Pass Women in Technology group.  I’m incredibly pleased to be sharing my professional and personal journey in tech, The Power in the Simple Act of Doing.

October 27th:

Yes, there is an Oracle event in the midst of all those SQL Server ones!  I’ll be doing three sessions at the fall Indiana Oracle User Group event!   I just sent over the abstracts after I let them know months ago I would love to speak with them again and it will include a couple new talks, including an updated Oak Table talk, Farming Dinosaurs and Empowering Data Access in the Cloud.

October 30-Nov. 3rd

Microsoft Pass Summit is HERE!!  This is annual geeky conference for the SQL Server experts.  I attended my first event last year and made it a goal to present the next year.  Well, after six months of presenting at SQL Server events/webinars, there I be!!  This is an incredibly welcoming community and guess what I’ll be speaking on?  DevOps Tools for a Winning Combination 

The event occurs over the Halloween holiday and due to this, I’m working on my costume, which will be as “dry” as my sense of humor.  I pack light, so the idea of how I could dress up and still only bring a tiny suitcase, (I hate anything more than a 20inch international carry-on.)  I figured it out.  I smart… </p />

    	  	<div class=

Indexing Seminar: Why Small Classes? (A Small Plot Of Land)

Firstly, a huge thank-you to everyone who has enroled in one of my “Indexing Internals and Best Practices” Seminars running soon is Australia. The feedback and overall reaction has been extremely positive and I can’t wait to present all the material I’ve been putting together these past few weeks. There are currently limited places still […]

Driving for Change- Women of Oracle

Somehow I’ve made it a pastime of pointing out how many times women are passed up for recognition while men are promoted around them. Its not that the men were identified as better or worse than women, but just about how heavily we scrutinize and limit women while everyone else passes them by. Now I’m going to ask the Oracle folks to not take this personally, but hopefully take it as a call to arms, but I know how we love to shoot the messenger.  Hopefully you already know that I will have no qualms about deleting abusive or threatening comments to this post, so please think twice before doing and really think on what I’m posting here.

As my technical work spans a number of communities, I have a broad range of experiences when it comes to how diversity is approached in different technical industries and its surprising how different each do approach it. It’s a complex challenge and of all the communities, (no matter if we talk Database, Big Data, DevOps, Testing, etc.) I have to give a lot of credit to the Microsoft SQL Server community on how they approach theirs as the most successful when it comes to women in the space. As with most community efforts, it’s driven by the people involved and the focus on the SQL community being one of family and treating each other with honesty and respect. It’s been an eye opener to me, just as it was when I adopted much of what I experienced in my limited involvement with them back in 2011 into my WIT mentoring initiative.

The SQL Server community has a healthy involvement of women in their MVP program, the peer recognition program awarded yearly on community and technical contributions. If you question the validity of this, just look at my twitter followers and count the number of women from the SQL Server community that are active and realize this: I have less than 20% followers that are women and I’ve only been speaking at SQL Saturdays for six months. I’ve been part of the Oracle community for 7 years and is one of the main women identified with women in technology when discussing Oracle. The difference in quantity of active women from the SQL Server community vs. Oracle is evident and it’s not just about being users or Twitter.

This brings me to this week, when the ACE Program, the comparable program from Oracle, published the official ACE Director photo:

https://i2.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 300w, https://i2.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 768w, https://i2.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 1312w" sizes="(max-width: 600px) 100vw, 600px" data-recalc-dims="1" />

My first thought was, “Where are all the women?”  There’s never been a huge group of women that have been able to meet the demands of the ACE program and I knew there were a few obviously missing from the program photo, but I was alarmed and decided to go to the website to see if the photo was just a misrepresentation of reality.

The Numbers

Now, the “find an ACE” webiste doesn’t have a gender filter for searches, so I had to do the best I could, going by name and by any photos if I was unsure. I knew what the numbers were when I was previously involved in the program from WIT presentations, so I used this as a comparison to the current program.

The current counts per the website are:

384 active ACEs, 25 women in the program


When I was part of the program, there were:

485 ACEs, 35 women in the program


The ACE Director Program is doing a bit better:

Directors total: 106, 8 women


When I was part of the program, the percentage of women directors was higher too, though:

112 ACE Directors, 11 women


With all the women in tech initiatives and programs since 2014, the numbers shouldn’t have gone down. I want to assure you, the ACE program is only one small area of Oracle and the program should be recognized for taking steps to bring more women in- I see them doing WIT luncheons and trying to get more activity around the women in the community.  The photo from Oracle Open World WIT ACE’s event demonstrates they’re really trying:

https://i1.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 300w, https://i1.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 768w, https://i1.wp.com/dbakevlar.com/wp-content/uploads/2017/10/Screen-Shot-2... 1312w" sizes="(max-width: 600px) 100vw, 600px" data-recalc-dims="1" />

The reality we have to face-  this is an event with what?  35,000 attendees and this is how many women recognized and in attendance?  I see ad after ad for Oracle Open World and among the shining faces beaming back, there are a significant percentage of women in each one, but when it comes to those contributing, speaking and in attendance, (I’m not talking about in the booths, sales or marketing, but at the deeper, technical level) we’re still missing. At this time, we should be past just luncheons and recognize that we need more than some food to fix what’s broke.

The simple truth is, we have a significant problem dealing with women’s ambition.

You Want Too Much

I have watched us fail the women of the community over and over again. We make excuses and scrutinize their contributions to a level that while they are wasting their time trying to convince everyone they are worthy, those around them, even those that may only be mediocre, are passing them by. After consistent roadblock to achieving the same as their male peers, they finally give up and decide that what contributions they make that are accepted are going to have to be enough.

We are responsible for this, me included. I’ve been advised not to back women’s inclusion in the ACE program, telling me that if I do, it will just come off as me backing the candidate because she’s a woman. I’ve been told excuse after excuse of why a woman isn’t qualified and then I’m left pondering as men, with fewer qualifications, are awarded tiers above and beyond the “lofty goals she aspires to”.

After Oracle, I aspired to have my ACE Director reinstated. I was asked multiple times “Why do you need to have your ACED back, Kellyn?” I was told, numerous variations of, “you shouldn’t just think you deserve to have it reinstated.” I just brushed it off, but the comments I received were incredibly alien to many of my male mentors. They couldn’t figure out why anyone would question my ambition to succeed or have something reinstated that I was so obviously still accomplishing. The bias that was shown to me by these individuals was beyond them and they called it as they saw it. I also knew that other women would be hearing similar statements when they attempted to accomplish what they saw their male peers doing.

Think Before We Advise

Before we offer advice on a female peer’s advancement opportunity or make a decision that could impact the opportunity, please consider the following:

  1. Am I offering honest and unbiased scrutiny of the female candidate?
  2. Even if I think I’m unbiased, am I experiencing any nagging feelings about the situation that I need to think through before I speak? This hesitation may be more about my own cultural upbringing on women’s ambition and not the candidate’s qualifications.
  3. Are their male peers to compare against and identify any bias that I may not have been first aware of?

So many are surprised when they take a bias test and find out just how much culture can impact us when making decisions and offering advice. There are a significant amount of tiers of bias that we have to face around women’s ambition and if we don’t, we will continually fail the women in the Oracle community.

The Forgotten

As I researched the numbers, I noted who had ascended to ACE Director status and I had no issue with them achieving the tier they had. I have always been for an open community- one that welcomes and promotes contribution, but forgive me as I say this, I was left noting that there are women still not ACE Directors that should have been a very long time ago and I’m going to discuss two of them here.

Michelle Kolbe requires no introduction. She holds two board positions, both on a regional and an independent user group board. She speaks at events, contributes content and has offered an incredible amount of her time to the Oracle community. I’ve watched excuse after excuse not to make her an ACE Director and [apologies to her, she deserves better than this…] it needs to simply be corrected. She is the epitome of what you desire in the Oracle community, yet to officially recognize this seems beyond us as a community.

Another one is Janis Griffin. I speak with her at both Oracle as well as SQL Server events. She has provided content, community support and valuable introduction to Oracle to those outside of our community. She’s been part of the community for decades, serving with user groups from the very beginning, yet she still isn’t an ACE Director. This is another failure for us to support, sponsor and lift up the women in our community.

Both of these women have been held stagnant at a status of ACE for too long and if you question it, compare them to their male ACED counterparts and think again. We are simply too comfortable reaching for our token women of the community and seem to be satisfied.

It’s an incredible group and I completely understand why people reach out to us.  Included with me are Debra Lilley, Heli Helskyaho, Edelweiss KammermannMaria Colgan, Michelle Malcher and a few others. If you were to try to list the main male speakers in the circuit, the list would vary greatly from person to person, but to name the women, it often comes back to the ones listed here when someone is searching for a woman to be included in an Oracle conference, webinar, book or interview. How do I know this? As I started out the paragraph, I’m one of this group.

In hindsight, I recognize that I created my mentoring groups and sponsored women with the plethora of opportunities that I was presented with as one of the “default” token women people reached out to. Our women in the community are rock stars, but we need to band together and start to help lift up those who aren’t recognized like we are. We need to help build them up, as they are invisible so often to the community as a whole.

Make a Change

I want to be clear on this- it falls to all of us to make this change. We women have a lot of power to do this for the community, but everyone is essential if you don’t wish to continue to see the numbers decrease. I wouldn’t be where I am today without Tim Gorman stepping up a very long time ago and telling me what I was capable of accomplishing. I wouldn’t have the career I have now if it wasn’t for Alex Gorbachev and Kerry Osborne saying, “come work for me.” I wouldn’t have received my first book deal if Jonathan Gennick hadn’t asked me to write on a SQL Server book. I wouldn’t have done as well with my time at Oracle if Mary Melgaard and a few others hadn’t gone through my offers and guided me to make the best decisions. Women need mentors and sponsors, along with representation. The community, especially we women in the more recognized group, should be reaching out to other women, offering to mentor and sponsor them.

We women need to stop thinking the only women we can mentor or sponsor are those that share our ideals. We have to stop letting people peg us against each other and follow their recommendations, (I’m at fault for this, too!) and if a woman is doing the job, making the contributions, we should then make a professional decision to back her. This is professional, not personal and we need to learn to separate the two, because right now, most of us are failing at this.

There is more than one seat at the table and the only way to address the “reverse quota”, (i.e. “we already have a female speaker” or “We already have a woman on the panel”) is to add more women representing Oracle. The more of us that there are, the more we shift from the “token woman” to being viewed as another tech expert.

The List

I’m constantly told that there aren’t women to speak at events, contribute written content or to take on initiatives. I say that’s BS and to make it easier for everyone, I’ve collected a list of those women I can think of right off the top of my head, along with links to their twitter profiles.

Mia Urman
Komal Goyal
Apurva Panidhar
Natalie Delemar
Karen Cannell
Courtney Llamas
Jackie Mailroy
Neviana Zhgaba
Paula Pullas
Flora B.
Carina Mendes
Opal Alaphat
Ann-Sofie Vikstrom Often
Ann Sjökvist
Rita Nunez
Sandra Flores
Sabine Heimsath

If you know of a woman I’ve missed from this list, please email me at dbakevlar at gmail and I will update this post regularly.

I’m through with what I’ve observed for the last 1 ½ years and no amount of excuses are going to fix it. Oracle, we’re better than this. We can do better than this.







Copyright © DBA Kevlar [Driving for Change- Women of Oracle], All Right Reserved. 2017.

The post Driving for Change- Women of Oracle appeared first on DBA Kevlar.


ODA X6-2

So, the ODA X6 End of Life seems to be planned for October, and be careful, as far as I know, this concerns also the storage expansion.

The main feedback from the X6-2 series were:

  • not enough cores when consolidating multiple databases in M, L and HA
  • not enough storage on HA when compared with X5-2
  • Standard Edition on HA would be great for cheaper RAC
  • The position of the ‘Large’ version of ODA Lite was not clear


The Small version of the ODA Lite has the same number of cores as in X6-2: 10 cores in one socket. Those are Intel Xeon Silver 4114 processor (2.2GHz). Some more memory: 6x32GB instead of 4x32GB, so 192GB expandable to 384GB as before.

And the big new thing is that it can run virtualized (X6-2 was Bare Metal only). Even if it is still OVM, the hypervisor is KVM. OVM is accepted by Oracle to license per vCPU, but for the moment the database cannot run on a VM so you need to get database licences for all the cores enabled on the ODA.

The storage for database is two 6.4TB Flash NVMe, which means 12.8TB raw capacity, 4.8TB double-mirrored, 3.2TB triple-mirrored.

ODA X7-2 M

You can look at it in the 3D view

The larger version of ODA Lite has more cores: 36 cores instead of 20 before: two sockets with Intel Xeon Gold 6140 processor (2.3GHz). Some more memory: 12x32GB=384GB instead of 256GB. Expandable to 768GB as before.

As the Small one, it can be virtualized with KVM.

The storage for database is two 6.4TB Flash NVMe, and you can add 6 additional ones to bring the storage to 51.2TB raw capacity.

Because of this new capacity in the ‘Medium’ one, and the large possibility to add disks, there is no ‘Large’ version of ODA X7-2.


You can look at it in the 3D view

The HA version of ODA has more cores: 72 cores instead of 40 before: two 18-cores sockets with Intel Xeon Gold 6140 processor (2.3GHz), in two nodes. Some more memory: 768GB instead of 256GB. Expandable to 1.5TB as before.

The big new thing is that you are allowed to run Standard Edition, and then RAC with no option. In the latter case, with SE2 only one socket per node is enabled.

The second important thing is that the storage which was limited (12TB expandable to 24TB or 48TB with a second shelf) is now declined into:

  • High Performance 3.2GB SSD for 16TB (raw, which is 8TB in normal redundancy or 5.3 TB in normal redundancy) expandable to 64TB, or 128TB with additional shelf
  • High Capacity starts the same with 4 SSD, but you add mechanical 10TB disks for additional 150TB to 300TB

For the REDO, there are 4 additional 800GB disks (4 additional in the expansion shelf).

In the initial release of X7 the virtualization on HA still used OVM/Xen.

Common new features

You will be able to install 12cR2 (

You have the choice for the network between:

  • two 10 GbE ports (RJ45)
  • or two 10/25 GbE ports (SFP128)

OAKCLI will be replaced everywhere by ODACLI



Cet article ODA X7 est apparu en premier sur Blog dbi services.


Here’s a quick quiz.

According to the Oracle 12.1 Database SQL Tuning Guide the first stage of parsing a statement is the Syntax Check, which is followed by the Semantic Check, followed by the Shared Pool Check. So where you do think the statement text will be while the Syntax Check is going on ?







And the answer looks like ….




… the shared pool. Here’s a simple test, cut-n-paste from SQL*Plus running under in the SYS schema (I’ve also done this in the past with older versions):




SQL> select user frrom dual;
select user frrom dual
ERROR at line 1:
ORA-00923: FROM keyword not found where expected

SQL> select kglhdpar, kglhdadr, kglobt03, kglnaobj from x$kglob where kglnaobj like '%frrom%' and kglnaobj not like '%kgl%';

---------------- ---------------- ------------- --------------------------------------------------------------------------------
00000000D1EE3880 00000000AEF43F40 bshhvh0ypcz6x select user frrom du
00000000D1EE3880 00000000D1EE3880 bshhvh0ypcz6x select user frrom du

2 rows selected.


So the statement is garbage and fails (or ought to be failing) on a syntax text – but I can find it in x$kglob – the library cache objects – in the SGA with a parent and child entry.

I have to say that when I first read, many years ago, that there was a syntax check it struck me that the fastest way of doing a syntax check on a “good” system would be to start with a search for the text in the library cache. If it were there (and on a good system it probably would be within a few minutes of startup) then you could have a flag that avoided the CPU cost of doing the syntax check and move straight on to the semantic (basically object and permissions) check.


Snapshots from OpenWorld so far

Bridge Run on Saturday






Swim in the bay on Monday




And the technical content begins !





Oracle 18c RU and RUR for pioneers and followers

In a previous post, I explained the announcement of the new release model, with Release Updates and Release Update Revisions replacing Proactive Bundle Patch and PSUs. Here is a description that will help you to plan your patching activity.

Which DBA are you?

There are currently a lot of discussions about the different types of DBAs. When it comes to patching, the DBA type is important.

Some DBAs are very conservative: do not use new features, do not upgrade to ‘.1′ releases, because every change can break anything. At least, maybe just patch to SPU for the security fixes. Some are more confident and apply the PSUs, with more fixes. The problem is that what is not fixed may be subject to a bug one day, and need an urgent one-of patch.

And some others are very pro-active: they patch and upgrade the development environment as soon as there is a new Proactive Bundle Patch, then they always have the latest fixes.

Actually, this does not depend only on the personality of the DBA but also on the ability to test. If you have only the production database with representative volume and activity, then any patch set can be dangerous. If in addition to that any intervention or decision takes 3 months, the problems encountered are even worse. But if you are in DevOps, where developers work on a production copy, when integration tests run every night, covering most of the critical use cases, then you can be confident that once those databases have run a few days with the latest Proactive Bundle Patch, then you can do the same in production. And if you come upon the low probability of a regression, you can react immediately with workarounds because you are agile.

The SPUer, the PSUer and the PBPer

In the previous release model, still used for 11g and 12c up to 12.2, this context will define if you are a SPUer (only security patches with very low risk of regression), PSUer (the critical fixes), or Proactive BPer (more fixes) but this will still not include optimizer fixes, and then may still require one-offs.

The .0, the .1 and the .2

With the new release model, you will have releases (R) every year, like 18c, increasing the first number, such as 18.1.0 which should arrive in January. Then during the support window, you will have Release Updates (RU) release quarterly to proactively bring software to the lasted fixes, increasing the second number, such as 18.2.0. But, because those may bring some regressions, you will have additional Release Update Revisions (RUR) which will fix the issues encountered in the last 6 months Release Updates, increasing the third number, such as 18.2.1 and 18.2.2

Then, there will be 3 types of approaches

The pioneer will go to the latest software

The pioneer is confident in his ability to cover most of the critical uses cases in his automated regression tests and is agile enough to fix quickly any issue in the non-covered cases. Then, he will upgrade to release 18.1.0 in January 2018. He will apply the Release Updates as soon as they are there: 18.2.0 in April, 18.3.0 in July, 18.4.0 in October. In January 2019 he can apply the 18c release update 18.5.0 or upgrade to 19c release 19.1.0

The wise will be on the latest Release Updates (RU)

Because each release brings a bunch of new features, they may come with regression on existing ones. Then, the wise DBA will prefer to wait that the pioneers have encountered them and that Oracle has fixed them in Release Updates. He can upgrade to 18.2.0 in April 2018, then apply the RU as they are released: 18.3.0 in July, 18.4.0 in October, 18.5.0 in January 2019,… He can continue for few years according to the support lifecycle (MOS 742060.1)

Of course, he doesn’t need to apply the RU immediately and can choose to lag for a few ones in order to be sure that more pioneers or eager wise ones have encountered more issues. But then, he will also lag on the security fixes, which is not good. And this is where Release Update Revisions come up.

The follower will be on the latest revision (RUR)

When you want to be up-to-date only on security fixes, with the minimal risk of regression, you will stay on the RUR branch, where the potential issues encountered on RU will be fixed. This is the same philosophy as applying only the SPU in the previous release model, but without lagging for years on other fixes. In the new release model, the RUR will be provided only for the last two previous RU. This extends the lifetime of RU by 6 months. The follower can apply the latest security fixes, but they will also come with other fixes. However, the risk is lowered here because the other fixes included are only those that have been tested for 6 months by other users. This 6 month lag on pro-active fixes keeps the risk minimal without running old software. He will upgrade to 18.2.1 in July for a 3 months lag, or waits October to upgrade to 18.3.2 for a 6 months lag.

So what?

Of course, you don’t have upgrade to 18c immediately and to patch every 3 months. But when you want to patch, you have the choice to be on the latest software to get new features immediately, or on latest RU to get new fixes immediately, or on latest RUR to let the others endure early problems for a few months. Be aware that RUs have a bit more than what Proactive BP included. In RU you may have some optimizer fixes, but those are disabled by default if there is a risk of plan change. You may also find some minor enhancements. You have the flexibility to switch between RU and RUR when you want, depending if you want some latest fixes or not. And in the low probability that you need a one-off patch, it will be delivered quickly for the latest RU and RUR.

Note that each release will be supported 3 years which means that you have the choice to go to the next one when you want the features that come with it. In 3 years, you will have the choice to run version 18.12 or 19.5 or 20.1 and with additional RU or RUR.

Just a guess: if Oracle continues with’Cloud First’ release, there are good chances that we will get 18c on-premises starting with 18.2 after a few months running 18.1 on the Cloud.

This new model helps to run a software that is not too old, which is more and more mandatory, at least for security reasons. Finally, all is about testing. Do you prefer to let others test the quarterly upgrades on their database, in the hope that what they run is similar to what you run on your database? Then be a ‘.1′ or ‘.2′ RUR follower. Or improve and automate your own testing and you can reduce the lag by being an up-to-date ‘.0′ RU patcher.


Cet article Oracle 18c RU and RUR for pioneers and followers est apparu en premier sur Blog dbi services.