Oracle Launches Globally Distributed Autonomous Database

Oracle recently announced significant updates to its Autonomous Database, enabling specific enterprise capabilities for data sovereignty, redundancy, scalability and AI-enabled usability that are novel in this market. This service will be of particular interest for organizations with a multinational footprint and multinational data-sovereignty concerns.

In the following sections, I’ll dig deeper into Oracle’s announcement and explain why this service could yield significant cost savings and operational improvements for IT organizations.

The Data Challenge For Multinational Operations

While data has been the fuel that drives the modern business, it is also a major challenge that enterprise IT must account for. We all understand the criticality—and complexity—of data privacy and integrity issues. However, we don’t all understand how much more complex and costly this becomes at a multinational level. Local laws and regulations around data sovereignty, privacy and survivability are just that: local. The nuanced language and requirements that apply from one country to the next require so much effort, attention and (local) resources to manage.

In fact, to adhere to localized data requirements, organizations will hire entire teams of data professionals—legal, operational and technical—to ensure adherence to both internal and external requirements. Additionally, entire infrastructure deployments are often used to guarantee that there is no daylight between what is required by regulations and what is delivered by the responsible organization.

Why do companies invest so much time and effort in adhering to local data privacy and sovereignty rules? Because not adhering is bad business. Pretty simple, right? Failing to adequately protect customer and company data is just wrong. It impacts the business both directly and indirectly, including through fines, bad press, lower stock prices, revenue loss and so on. As one example among many, in 2021 Amazon was fined roughly $840 million for violating the European Union’s General Data Protection Regulation.

Achieving Distributed Database Performance With Sharding

To achieve data sovereignty, many multinational companies have employed database sharding. In simple terms, this no-share architecture uses a partitioning process by which a database is essentially carved into smaller and separate parts. Each shard is a physical database and runs independently of other shards (databases) in this environment. So, the data in India is unique to India and stays in India, while that shard is still connected to the larger global database environment and the data can be accessed from anywhere. Why is this important? It enables organizations to set and enforce policies and practices around data privacy, sovereignty and the like on a shard-by-shard—that is, country-by-country—basis.

While sharding is an effective way to build a distributed database environment, there can be downsides. The first is application performance, or latency. If an application request originating in the U.S. calls on data from a shard in India, it will require more time for that data to be retrieved. For a local request, performance will be better.

The second downside is complexity, which can impact scale. As more shards are created, the amount of cross-shard SQL increases considerably. This creates more overhead and can, at some point, become a detriment to performance and manageability.

Oracle Globally Distributed Autonomous Database—Sharding And More

Sharding is essential to the distributed nature of Oracle’s Globally Distributed Autonomous Database, which operates as one big logical database that maps to multiple physical databases (shards) where local data is stored. Each shard is replicated for survivability, and shards run in an active-active mode, meaning that all shards process application requests.

Any database professional who has used or managed Oracle’s Real Applications Clusters database understands that the scale-out capabilities developed for that clustered database deliver top-notch performance. For its implementation of the new distributed database, Oracle built on top of local RAC and Autonomous Database capabilities so that each shard can be easily and independently scaled up to achieve high performance, and then scaled back down to manage cloud consumption costs. The result is a distributed database that enjoys the scale, redundancy and sovereignty associated with sharding combined with the performance and local fault tolerance you would expect from the RAC and Autonomous Database products.

When combined with application-transparent access to the sharded database, these capabilities enable applications to run with high relational database performance and without changes. In short, Oracle’s Globally Distributed Autonomous Database helps customers reduce complexity while addressing their data residency, performance and availability goals.

This is also important as it differentiates Oracle’s offering from many of the distributed database offerings on the market, as those competing offerings are built on NoSQL. Oracle’s approach capitalizes on its decades of experience delivering high-performance, mission-critical relational databases that meet ACID requirements for full consistency. The NoSQL approach, by contrast, introduces a number of data consistency challenges as well as providing reporting and analytics that simply fall flat in a distributed deployment. In fact, many NoSQL vendors appear to be trying to add SQL on top of a NoSQL engine, while others are trying to support distributed scale-out. Neither of these fixes is trivial, and they will probably take years to achieve.

One of the keys to effective distributed database environments is data distribution across shards. This is another area where Oracle has differentiated itself from the competition by enabling different distribution methods based on each organization’s needs. Distribution is also managed automatically based on policies defined by the user. For example, organizations may want to distribute data based on country of origin to address data residency concerns, use a hash distribution for scalability and parallelism, employ user-defined distributions to address data-specific skews or even replicate tables that are heavily read-oriented to every shard for the sake of efficient database operations. In fact, Oracle allows customers to flexibly combine these and other approaches to match their specific needs.

Additionally, Globally Distributed Autonomous Database offers the widest variety of replication methods—synchronous, asynchronous, adaptive synchronous or a combination. Finally, Oracle seems to have analyzed every option when considering how and where shards are deployed—regular servers, fault-tolerant scalable clusters, in the cloud or multiple clouds. All of these options give organizations maximum flexibility for deployment and use of Globally Distributed Autonomous Database.

Putting The Autonomous In Globally Distributed Database

With all of the work that Oracle has put into its Globally Distributed Database product over the years, the big news is that everything described here is available as an autonomous cloud service in Oracle Cloud Infrastructure as well as through on-premises deployment via Exadata Cloud@Customer or Dedicated Region. Each shard is treated as an Autonomous Database by adding Globally Distributed Database capabilities to Autonomous Database. This means that each shard is elastic, auto-managed, auto-scaled, auto-tuned and auto-patched. While it may sound overly simplistic to say that organizations can “deploy and forget” with this service, practically speaking it’s true.

When all of the capabilities outlined above in terms of scale, data provisioning and the like are combined with the capabilities of Oracle’s Autonomous Database in OCI, it creates a distributed database solution unlike anything else on the market.

One of the very cool capabilities that Autonomous Database brings into the equation is what Oracle calls Select AI. This tool essentially converts simple text questions into SQL statements for execution and enables conversational threads. In Globally Distributed Autonomous Database, the query is automatically routed to the appropriate country or shard to generate the answers. This is a powerful capability for organizations focused on enabling business users, because users no longer need to know either how their data is laid out or where it’s physically located.

As somebody who learned and used SQL*Plus and PL/SQL, this kind of feels wrong as it takes away the power of being a nerd. But in all seriousness, Select AI is a stroke of genius that empowers all users to make the best use of data.

What This Means For Enterprise IT

It is a little too simplistic to say that deploying your global data estate with Distributed Global Autonomous Database is as easy as pointing and clicking. However, it does make it far simpler to navigate the ever-changing regulatory landscape of individual countries and manage those individual data environments (shards). What once required a team of database specialists working in concert with legal and operational experts can now be performed centrally and from a single console.

As an ex-IT leader who had to manage a team of database specialists for dozens of state agencies, I fully understand the organizational and operational challenges that appear on the radar day after day. Adding the challenges associated with sovereignty, redundancy and performance across a number of countries? Forget it. Faced with that reality, my first order of business would be reducing cost and complexity—and that would lead me to look for a solution like Globally Distributed Autonomous Database.

Closing Thoughts

Oracle’s Globally Distributed Database delivers a breadth and depth of functionality that can come only from servicing the data needs of the enterprise for decades. This is what makes this solution stand out in the market: it’s the understanding of how data is used and managed around the globe that translates into features and capabilities that other vendors simply haven’t thought of or achieved.

By enabling this as an Autonomous Database service, the company further addresses what enterprise organizations want and need as their data estates grow to span the globe. Add in new AI-enabled capabilities, and this becomes a modernized data platform for the modern organization running modern applications.

I’ll be keeping a close eye on how Globally Distributed Autonomous Database lands with customers, so check here soon for more insights.