Key Takeaways from Oracle CloudWorld 2022

I have attended and spoken at various Oracle user conferences since the 1990s. Every few years, Oracle shakes things up and the conference name evolves. The 2022 event also saw Oracle rename its conference from OpenWorld to CloudWorld. But this time, it was not merely a rebranding exercise, but it suggested something bigger.

Oracle CloudWorld 2022 demonstrated a new level of confidence in Oracle’s embrace of cloud computing. Its Oracle Cloud Infrastructure (OCI) is rapidly expanding its global footprint of data centers and innovating rapidly. It has gone from 1 data center in 2016 to 41 regions by the end of 2022. Cloud is expected to be 30% of its revenue according to this excellent presentation.

One of the biggest themes of the conference was the multi cloud aspect of OCI. OCI has been a trendsetter in minimizing egress costs from its cloud. Hopefully, other hyper scalers will also follow. The announcements ranged from MYSQL on AWS to Oracle Database Service for Azure. The latter allows Azure users to connect to Autonomous Database (ADB), Exadata and MySQL Heatwave.

This blog covers some of the vast number of announcements in the data and analytics space at the conference.

Oracle 23c beta launch

I have been a practitioner of Oracle databases since release 5 in the late 1980s. Yes, I am showing my age and have seen major evolutions of Oracle’s namesake flagship product. This year showed no letup in the company’s innovation engine.

Oracle’s database strategy is predicated on two principles — converged and autonomous.

Oracle’s converged database approach is to add multi-model capabilities right inside their database instead of offering specialized standalone databases. For example, Oracle provides blockchain capabilities, but as an immutable cryptographically signed table type. While there is no black & white approach that says one approach is better than the other, having fewer databases decreases product sprawl, reduces integration overhead, and minimizes security exposure.

Oracle’s Autonomous Database (ADB) builds upon its Real Application Cluster (RAC) parallel offering on its Exadata infrastructure. Amongst 300 new features and enhancements (over innovation release 21c), is its Oracle Database Zero Data Loss Autonomous Recovery Service. This fully managed OCI service is an analog to Oracle Zero Data Loss Recovery Appliance which runs on-premises. It automates continuous (aka real-time) data protection with predictable point-in-time-recovery at low cost and with protection for ransomware.

One of the exciting new features is JSON Relational duality (see below). Since Oracle follows the converged approach, it provides JSON capabilities right inside its single database. However, in the cloud, the Oracle markets its core ADB differently depending upon the use case. Hence, there are offerings like Autonomous Data Warehouse (ADW), Autonomous Transaction (ATP), and Autonomous JSON.

JSON Relational Duality (Beta)

Handling JSON documents in relational databases has been a conundrum. In the very first iteration, relational databases just stored JSON documents as a BLOB column and then flattened JSON hierarchies into a single column. However, that limited JSON’s semi structured hierarchical benefits. In the next iteration, databases added a native JSON data type that allowed documents to be stored in a more natural manner. Oracle’s Autonomous JSON supports this. Today, with duality, Oracle in effect encodes the hierarchies so they can be “reinflated.”

So, what is this duality?

In release 23c, Oracle puts a twist on its JSON story by providing a relational storage engine but access that can be JSON or SQL. This allows Oracle to address the impedance mismatch between application developers and the data modelers, while still benefiting from all its database optimizations and the security model of its RDBMS. Oracle tests show they get better performance than the traditional object relational mapping (ORM).

Oracle’s Relational JSON duality means that if an attribute value changes, that change doesn’t have to be made to every single impacted JSON document, but it can use the power of foreign keys to make that atomic change instantly in the schema.

Along with the JSON extensions to SQL, Oracle’s other access options include MongoDB API compatibility and its own Simple Oracle Document Access (SODA) API. Oracle knows it doesn’t have 100% MongoDB compatibility, but it feels it addresses 80% of MongoDB apps and is closing the gap as customers request the missing functionality.

Operational Property Graph

The plot thickens… Oracle has also added “converged” graph capabilities. Up to now, Oracle has offered a graph database, but it’s a standalone one. In release 23c, this graph converges right inside the overall database. But there is more to it. First, this database supports operational real-time use cases. More interestingly is the support for a new standard called SQL/PGQL. Property Graph Query Language is an open source extension to SQL.

This option supports microservices SAGA pattern through lock-free column reservation.

Multi Cloud Support

Perhaps the biggest change one can witness in Oracle’s cloud strategy is its embrace of multi cloud by expanding to Amazon Web Services (AWS). Before we talk about AWS, let’s revisit its Azure Interconnect which launched 3 years ago.

Oracle has expanded its Microsoft interconnect partnership which provides under 2ms latency between the two providers’ clouds. When Oracle announced Interconnect for Azure in 2019, it required heavier setup. In July 2022, Oracle simplified the process of connecting Azure applications with Oracle databases on OCI. And with no ingress and egress fees.

At Oracle OpenWorld 2019, I had connected with Mestec Ltd, a UK based company, that had the honor of being one of the first ones to put the Azure Interconnect into production. Now, three years later, they have expanded their use of Azure Interconnect. It turns out that 12 of Oracle’s 41 regions are close enough to Azure’s and so the Interconnect capability has been well received by customers wanting to run their .Net apps in Azure with the backend database in OCI at a lower cost and taking advantage of OCI’s native database optimizations.

Oracle is so bullish about the Interconnect’s success, that I believe it will expand this to AWS shortly.

Now, comes the interesting part, Oracle has expanded its MySQL HeatWave footprint to AWS. But, unlike the Azure offering that has the data plane in OCI, the new release’s data plane, control plane, and interactive console reside in AWS.

Since we are discussing MySQL, there are more interesting enhancements in its HeatWave offering. Read on.

MySQL HeatWave Lakehouse (Beta)

The new beta release marks Oracle’s entry into the red hot lakehouse space.

MySQL HeatWave has been on a tear for the last two years as Oracle has added analytical, machine learning, and database management automation capabilities to the open-source operational MySQL. Until now, MySQL HeatWave could ingest data from MySQL’s InnoDB into its analytical engine. In the new release, MySQL HeatWave Lakehouse, it now supports reading data from object stores (in formats like CSV and Parquet).

The data is first compressed and then analyzed in memory. It should be pointed out that the first approach based on InnoDB supports read and write capabilities, while the object store option is read-only. Oracle added new MySQL Autopilot for Lakehouse that further reduces manual labor and improves performance: auto schema inference, adaptive data sampling, auto load, and adaptive data flow are some examples of the AutoPilot feature.

Customer Roundtables

I got to attend several customer sessions that ranged from Deutsche Bank to Lyft to a dozen others. Deutsche Bank talks about their journey of migrating 10K on-premises databases to OCI.

My findings from the roundtables:

1. Multi-cloud is becoming mainstream — Deutsche Bank collocation with GCP, Mestec Azure Oracle Interconnect are some examples.
2. ADB is mainstream with users like Scholastic, Mestec, Forth Smart, Lyft, and Outfront mentioning its auto scaling capabilities
3. ADB and MySQL really works with minimal DBAs — Outfront, Tetris
4. APEX is being used at enterprise scale — Munich RE, Outfront, NRI
5. Exadata benefits many types of customers. Some are at multi petabytes scale. Mestec like its use of ATP as a slice of Exadata without having to pay for full Exadata.
6. ADB and MySQL HeatWave enabling Machine Learning — Forth Smart, Tetris
7. Exadata Cloud@Customer is a great first step to cloud — Deutsche Bank