Full Stack Development Using Oracle 23ai

Hi. Today's presentation is about full stack development using the new Oracle 23 AI database. It also uses Parse Platform as the backend server.

Parse Platform is known as a MBaaS server or a mobile backend as a server and facilitates rapid development of mobile and web applications. The server runs against the new Oracle 23 AI release. This is a full featured, free version of the Oracle database running as a Docker container.

My name is Doug Drexel. I have worked in software for over 30 years. Maybe you know some products I've worked on, which include Tuxedo, WebLogic Server. I've worked on the Oracle Kubernetes engine for a bit. And now I'm working with the database team in developer outreach.

So very quickly, as you probably all know, full stack development refers to the practice of building and working on both the frontend client side and backend server side components of a mobile or web application. Full stack developers have a comprehensive understanding of how the different components of a web application interact and integrate with each other. They can work on the user interface, handle business logic, manage databases, and ensure seamless communication between the client and the server. The ability to work across the entire stack allows full stack developers to build complete web applications from start to finish without being limited to just frontend or backend. This versatility makes them valuable assets for companies developing web based products and services.

In this talk, the backend will be Parse Server. The database will be Local23AI, and the frontend will be a variety of SDKs that Parse Server supports. So, Parse Server is an open source Node.js express app that is commonly referred to as an Embedded Server. Oracle has developed a new storage adapter for Parse Server using the Oracle NodeDB modules. The database is the new Oracle-free 23AI running as a Docker container. This container supports a NoSQL development using customary JSON syntax. The frontend is the variety of SDKs that Parse Server supports. What is new and unique about this stack is this is the first time Oracle has offered a free database container for local development.

All right, let's talk a little bit about Parse Server. As a developer, you know that building mobile apps can be a complex and time-consuming process, especially when it comes to handling server-side logic and data management. That's where Parse Server comes in, offering a powerful and flexible backend solution that simplifies mobile development. With Parse Server, you can quickly set up a robust backend infrastructure, complete with features like data storage, user authentication, push notifications, and more. This allows you to focus on building the core functionality of your app without getting bogged down in the intricacies of server setup and maintenance.

Parse Server's open source nature also means that you have full control over your data and can easily customize the server to fit your specific needs. Whether you're working on a small personal project or a large-scale enterprise application, Parse Server can help you streamline your mobile development workflow, saving you valuable time and resources.

Here are the software development kits that Parse Server supports. I have used React Native for a walking history app that is going to be used in the workshop. I have used Flutter to create a simple Game Score app based on Game Score, which is Parse's Hello World. That app implements authentication and authorization and uses GraphQL internally. A team member has used the Unity API SDK, a common gaming SDK, to create a rock-paper-scissors gaming mobile app. The Parse Server API provides a comprehensive set of APIs and services that allow developers to build and manage the backend infrastructure for their applications. It includes the following components: data management, user management, push notifications, and cloud code to write custom server-side code.

Data management, user management, I have basically used all of these, push notifications, cloud code to write custom server-side code, and we will touch on this throughout this talk.

And that concludes a summary of Parse Server, the backend. Now to the database. All right, I am going to talk about the new 323AI Oracle Database Container. Yes, a Docker image for Oracle that is free. It is a full-featured database image. There is nothing missing here. To use it, you have to do a Docker run, and it will download the container and start it.

And then if you want to use it for JSON development, you have to run these commands to enable JSON support. To do this, you will have to download the SQL client to interact with the local container and then you will connect, see this command, connect to the image of sysdba. You are going to connect as the systemdba to a pluggable database called 3PDB1. You can see that in the SQL command. Once you log in, you will run these commands, or you can run them as a script. Just pipe them into SQL and that will set up the local Docker container for JSON development.

But what is a pluggable database? So Oracle has this idea or concept of a container database. And 323AI is a container database, which means there may be many instances of a pluggable database, which could correlate to cloud native services or bounded context. So for each PDB inside a container database, you can define that. Here, let me try this again. In a microservices architecture, a pluggable database can be scoped to a services bounded context. For example, the supplier service, the customer service, and the order service. While at the container level, the sysdba, you can still manage all the PDBs as the same as far as patching, upgrade and backup. Another big plus is that a team can collect and report on metrics for all PDBs in a container database.

We saw the sysdba administrator that could run commands at the container DB level. But every PDB can get its own administrator that maintains the specific instance. For the default PDB, free PDB1, the default administrator is PDBadmin.

A little bit about the storage adapter that we built at Oracle for Parse Server. Right now, the Oracle storage adapter code is hosted in the Oracle Samples GitHub, and it is recommended to work with the latest release of Parse Server 7.0.0. I am working with the Parse team to release this adapter as a, quote, proper adapter. Currently, the other storage adapters, Mongo and Postgres are part of the server source base.

Currently, the other storage adapters, Mongo and Postgres are part of the server source base. So that's what I mean by a proper adapter. We wanted to make storage adapters dynamically configurable. And I've talked with the Parse team, and they want to also move out Mongo and Postgres from the server repo and turn them into their own repositories. So to do that, I worked with the Parse team to add a couple of features, a test exclusion list, which allows each adapter to disable broken tests for that specific build. And I added some storage adapter support, which allows for the dynamic configuration of a storage adapter.

We also have a GitHub pipeline that runs the full test suite against Oracle. And we currently pass 2,671 tests out of 3,141. Of those failures, 253 are Oracle-specific as identified by the test exclusion list. I was talking with the Parse folks this morning, and we intend to make this adapter officially available by July 1st. And right now, it is supported as a developer preview. So when you install Parse Server, you also get a dashboard that's really simple to install and to work with your server.

This dashboard allows you to do a bunch of things. It allows you to manage users. It allows you to manage collections. You can configure webhooks to call other Kubernetes services, for example, services provided by our Spring Boot platform. We have a Spring Boot platform, and I've used webhooks to call the services offered by those Spring Boot microservices, and made them available through the API to mobile and web developers. And then, as you see down at the bottom of the console, there is the API console. You can see on dashboard start, if you take a look at that, I specified to connect to the Parse endpoint, which is the Parse API. I also specified to connect to the GraphQL endpoint, so we can actually run a little GraphQL through the playground. So here's a picture in the dashboard of the GraphQL playground. I found these playgrounds to be very useful as I ramped up on using the Parse APIs. You can write simple code that tests out the APIs, and you can verify that your code is working before you put it into your applications. So this dashboard has three playgrounds. It has the REST playground, the GraphQL playground, and the JavaScript playground. In this specific GraphQL example, this is creating a course document in the course collection. One of the things I really liked about GraphQL is its self-documenting feature.

It has the REST playground, the GraphQL playground, and the JavaScript playground. In this specific GraphQL example, this is creating a course document in the course collection. One of the things I really liked about GraphQL is its self-documenting feature.

I think that is neat. GraphQL's self-documentation capabilities allows developers to introspect and explore the schema and available queries and mutations, and it greatly enhances developer productivity to facilitate efficient collaboration across service teams. So that was sort of a quick, broad overview of this full-stack development environment and the different components. The backend Parse server, the database, Oracle 3.23 AI, and basically any SDK you prefer to use.

But now I'm going to start talking about things that I have done that are maybe Oracle-specific. I'm not sure, but these are things I've actually done with the stack. All right. There are a lot of acronyms in this title, so let's get through this. All right. Parse provides the idea of cloud code, which is server-side code that can be easily added to the client API set. Using cloud code, custom Oracle code can be executed. You can use any SQL, JSON, pre-existing PL SQL, or any of the queuing methods related to a queue, advanced queuing. In this use case, I configured a message queue and registered for a callback when new messages appeared on the queue, and that is the CQN, continuous query notification.

When a message was received to the queue, the callback would be called, and using APNs, Apple Push notification service, it would push the notification to users who subscribed. Quick shout out to Codeco because I used their Push notifications tutorial to build out this small application. This next example is similar to the first, except now I am working with Oracle Spatial. Oracle Spatial is a comprehensive set of features for managing and analyzing spatial data. In this case, I would similarly listen for events on the queue, but this time the events are related to trucking. Did the truck stray off course, as you can see? So this is sort of a geofencing application. If you look at the picture, we have given a boundary, a geofence boundary, and we can get alerts if a truck strays off course, if the truck is going to be arriving in 20 minutes, so we can prepare the bay. I have had some other ideas with this capability for some other things that I hope I am able to work on. So these are two things that I have done with this stack locally. I actually pushed messages, notifications through APNS off from my local machine, but there are so many other features to be explored in this release that I am looking forward to using. This is one of the next things I am going to be working on. These are property graphs. If you are not aware of what property graphs do, Oracle has a graphing database that you can do graph analysis on, and there are some examples of why you would want to do such a thing. What I want to do, or I am actually looking at, so we have this idea of a SQL tuning set.

Oracle has a SQL tuning set, which collects all the requests coming into the database and logs them. What I am doing is I am looking at converting a SQL tuning set to a graph and identifying related tables using community detection. This would be useful in defining a bounded context across existing tables by identifying joint affinities or maybe identifying transactional context.

So the work here that I am doing, and there are a lot of opportunities to use this feature in different ways, but the work I am doing is sort of migrating or taking standard SQL databases that have been around for a long time, looking at them and sort of helping our users to define bounded context so they can abstract the existing tables into services.

The next thing I am looking at, and this is in correlation with the SQL tuning set, this is called a JSON relational duality. What this means essentially is you can define JSON views on existing SQL tables. Using duality views, data is still stored in relational tables in a highly efficient normalized format, but is accessed by apps in the form of JSON documents.