Realtime processing of events at 100km/h - a tale of speed, saltwater and software architecture

A look into DRCP, a feature that could help with application servers that don’t support connection pooling. What is it? How does it work? Is that going to solve all your problems? What’s the impact on performance? Should you use it? How to use it?

In this session you will get the opportunity to experience Oracle’s latest Innovations in its MySQL-HeatWave technology. Using MySQL native interface, organizations can now achieve the most out of their data and enhance use-cases with innovative capabilities such as Realtime-Analytics, Lakehouse and Machine-Learning.

Oracle Database provides us with collection since a long as I can remember (Oracle 7.3.6.4). First as PL/SQL Tables in PL/SQL only. They have been been renamed (Index-by Tables, Associative arrays) and new implementations have been added since (Varrays, Nested Tables [SQL and PL/SQL]). Also the original Asscociative arrays have been updated, making them available in SQL as well. They play a key role in many of the newer Oracle features like BULK Processing (BULK COLLECT and FORALL) and Table functions. Using the bulk processing capabilities of your Oracle Database they can speed up your processing by an order of magnitude. This session will show a little bit of history, how to use the collections, how to make collections available in SQL so they can be processed in other languages. How things can be done in what version and how things become easier with the newer versions of the database.

So you have a major problem with your database. There’s a block-level problem. What can you do? First of all, call someone who knows how to fix your system. Don’t be an idiot and try to fix it yourself! If you don’t have the money to hire such a rare consultant, or are not fixing things but need to edit a block for other “reasons”, or you just want to understand the block structure a bit better. Come have a discussion about it. Let see what we can do to fix such problems. Lets see how rico2 works and what you can and cannot easily change within a block.

With highly automated deployments and higher deployment frequencies the need for staging non-prod databases is rising. During the years and Oracle versions the ways and possibilities for doing that increased. In this talk I will discuss some situations and ways for doing that and some ways I’ve seen in real life szenarios. Further I will provide some insights on discussions why some ways are not always workin

One of the nice little features in Oracle 23c is the “SQL Analysis Report”: When displaying execution plans, in some situations a report appears with suggestions on how to rewrite your SQL statement to improve performance. What is behind this feature? Did Oracle just added a list of fixed recommendations, or is an intelligent set of rules implemented to suggest specific performance improvements? How useful are these recommendations? Do they also work for complex SQL queries? There is no information (so far) about this feature in the Oracle documentation, but the interest of the presenter was definitely aroused. With various live examples I will show in which cases the “SQL Analysis Report” gives useful performance tips, how intelligent these recommendations are and where the limits of this new performance feature are.

DNV embarked on a transformation journey in 2020, migrating Oracle EBS ERP to OCI within a 7-month timeframe. We serve offices in more than 100 countries around the world with over 12000 users.

DNV has implemented an independently-developed solution called Dynamic Scaling which empowers them to always deliver the exact database performance to their EBS application as well as reducing cost significantly. Dynamic Scaling solution improves performance significantly, allowing clients to perform month-end closing 300% faster without any end-user manual intervention.

This session will cover DNV’s strategies to migrate customer to OCI, their real-world observations and metrics on the impact of its scaling tools and custom-tailored solutions, and a live demo of how simple it is to configure and implement Dynamic Scaling.

Graphs aren’t new in Oracle Database: RDF graphs are supported for a very long time. More recently, Oracle added support for property graphs as part of their converged database approach (it started before already, but was used as an argument of the converged theme). It was mainly through a separate tool (Oracle Graph Server) and query language (PGQL) that Oracle developed.

Now, with the Oracle Database 23c release there is support for the new SQL ISO standard covering property graphs: SQL/PGQ.

Is this one graph too many? Because it doesn’t replace anything, it is added next to the existing solutions. The most confusing part is that syntaxes are extremely similar (done on purpose), and users could be surprised to have errors on a syntax they found in the doc (but it was the syntax for the “other” tool).

In this session, I will first cover the main differences and pros and cons of the existing solution, and then also provide some hints about why it still does make sense that both solutions exist because they do have some key differences making one the best “tool for the job” based on the needs. While also covering the huge limitations SQL/PGQ has for now, making it pointless for many usages.

A practical presentation, covering SQL/PGQ, one of the 23c new features, and helping you in adopting it (or not) in the right way with some examples.

Data Pump is well known to almost every DBA. It is the Swiss army knife for all types of upgrades and migrations, regardless whether you migrate only to a new platform, a new version or the CDB architecture. Data Pump does it all for you.

But reality bites sometimes. There are useful Best Practices you may not be aware of. And in this session we try to show the best and well kept secrets. Why is there only one worker working on your 10TB LOB table? Why does it matter how many dumpfiles you use? Is network mode more efficient, less efficient, or about the same as using a dumpfile? How can you monitor what Data Pump is actually doing? And why is the Bundle Patch so incredibly important but not included in any RU yet?

We’ll have so many useful topics, it will be hard to fit them into 45 minutes. But we’ll get the best ones for you, and whether you are a casual user or an expert, you should learn something in this session you were not aware of before.

It was on a cold day in November when the silence was broken by a desperate scream: ‘Kill it! Kill it!”. The boss immediately pushed the button and brutally terminated the session. It has happened again… The Performance was murdered. Boss flicked through the case files. There were no usual suspects. Where is the murderer hiding then? The desk was buried under the cases’ files similar to this one. The File that hasn’t been seen in production for 6 months, the whole team has been looking for it, but there was no progress, not even a clue… Another reported missing - Data has left the Source System but never reached the Destination server…Nobody has seen it again… This is too much for one person. And that’s why I need YOUR help! During the session, we will solve all the mysterious cases together and we will find out what is responsible for the strange behavior of the system. Is the source of the problem a huge design error or a small decision at one of the stages of development? How minor design decisions can affect the system’s performance and vulnerability to errors? Who is the murderer? The answer may surprise you!

IPv6 is inevitable. We have known for a long time that IPv4 addresses are finite. And although the successor IPv6 has been available for twenty-seven years, most companies and all public clouds still work with IPv4.

The question is not whether we upgrade all our networks to IPv6 but when. It makes sense to gain experience with isolated IPv6 projects now. And the cloud is the ideal place for this.

This presentation introduces the differences between IPv4 and IPv6 and shows how to set up IPv6 networks in the OCI. I also dispel common prejudices against IPv6.

This presentation contains detailed live demos.

utPLSQL (UT3) - is a tool designed to carry out Unit Tests of applications written in PL/SQL. Practice shows that this tool can do much more than just run unit tests. As part of the presentation, the use of this tool in real projects will be discussed - from simple unit tests, through code coverage tests, to integration with CI/CD tools. In addition, the idea of using utPLSQL to carry out more and more sophisticated test scenarios, or to cover new functionalities or bugs with tests thealized with the use of this tool, will be presented. The real life scenario on which this presentation is based is a project on which several programmers worked simultaneously, in which several million lines of PL/SQL code were written and in which over 4,000 tests were created using the utPLSQL tool. Finally, the whole thing was fastened with the help of CI / CD framework, giving programmers an easy-to-use environment that minimizes the creation of code containing new errors or restoring errors previously detected and fixed.

I would like to present a couple of performance cased with my approach how to find the issue and fix it. The presentation will contain real scenarios.

Native support of JSON within Oracle Database became available in version 12.1.0.2 and has greatly expanded and improved in every version since. This session will give you an overview and examples of the various functionalities to marry JSON and Relational data across the versions from 12.1.0.2 until 23c including:

Different datatypes to store JSON
Discover content of schemaless JSON with Dataguide
Validate JSON with a JSON Schema
Querying JSON data in different ways
Creating JSON data in different ways
Modifying JSON data
Using JSON Relational Duality view to work with same data both as JSON and as relational
JSON in PL/SQL
GeoJSON and Spatial

Good things come in small packages. But is AI up to the challenge? A lot of AI examples are of image processing looking for images of cats and dogs. While somewhat helpful, these examples can be difficult to relate to. What about trying to find something useful? This session will be a demonstration of using Oracle’s AI Image Processing cloud service to process images of past Oracle Conferences, create an AI model, and use this to identify a small but very important part of the Oracle User Group Community. Can we find Martin Widlake. Just like the Finding Wally puzzles, finding Martin can be a challenge. Join us to see how we can solve this challenge.

The Power of SQL Trace and the data it contains is great, but often it is seen as complicated or even cumbersome. But with proper preparation it can show a great variety of information to DBAs, developers, operational teams and architects. The presentation shows why and when SQL Trace can help (and even it’s limits). The 2nd part is to understand methods to enable it, either from the application or DB side. Even in the cloud, different cloud providers provide their interfaces to enable SQL trace and get the tracefiles. And with 23c, there is a nice improvement also.

Since its brief pause in 2015, the DBMS market has reeled off 7 consecutive years of growth, with the last 6 years all growing in the teens. There are more than 400 database brands right now. The Database market has seen enormous change with the transition to cloud-based systems. Cloud platform providers lead the way and continue to make gains based on their performance. There is one database brand that not only stays dominant throughout the years but increases constantly it is significance among clients both on premises and in the cloud: The Converged Oracle Database.

Taking the high-altitude view there are only three reasons why you may be called to address a performance problem.

A specific task takes too long to complete A “batch” process recently over-ran its SLA “The system” is too slow

There are common elements to the three classes of problem, but the fastest solution to the a problem starts by identifying the correct strategy and using the right tools in the right way.

Ultimately many problems come down either to interpreting execution plans and know why the optimizer has picked a bad plan, or to code that needs to be re-engineered to eliminate single-row processing (or minimise its impact) or reduce concurrency; but the best way to find the bad plan, or identify the critical code depends on the basic nature of the problem and the way the cry for help appeared.

In this talk I’ll discuss the what, the how, and the why, of quickly identifying solutions to 99% of performance problems by doing the right thing at the right time.