Oracle OpenWorld 2018 - making the cloud friendly for enterprises that distrust cloud

Profile picture for user kmarko By Kurt Marko October 23, 2018
Kurt Marko adds his spin to Oracle's opening OpenWorld announcements.

As Derek du Preez noted, Oracle used the Open World event to showcase its continuing efforts to transform its eponymous database and other enterprise applications into cloud services.

Indeed, one look at the top of the company's website with rotating promotions of its self-styled Autonomous Database, i.e. a product using many now-common self-tuning and management automation techniques, illustrates how important the cloud, particularly database services, is to Oracle.

While the company’s Oracle Cloud portfolio spans the gamut of as-a-service capabilities with infrastructure, platform and software products, it naturally plays to its strength by emphasizing database services.

As such, this year’s spate of Open World announcements are full of enhancements to the Autonomous Database service introduced at last year’s event, including a new Exadata-based Autonomous Transaction Processing service and new security services that wrap around the entire portfolio (see our other OpenWorld cloud news here and here).

Oracle's cloud strategy reflects its database heritage and a customer base that has long relied on Oracle software to provide the foundation for critical enterprise business applications. Thus, it's understandable that Oracle emphasizes database services and claims the superiority of its cloud infrastructure for DB-backed legacy systems. In an interview with diginomica, Kyle York, VP and GM of Oracle Cloud Infrastructure, said its service portfolio is built upon the following principles:

  • IT investment protection, namely minimizing the need to redesign or update legacy software to run on cloud infrastructure.
  • Enterprise-capable infrastructure and services with the reliability, robustness, service level guarantees and security (more on that below) to operate critical applications responsible for revenue generation and core business processes
  • Price-performance advantage with services that deliver superior speed and throughput for the money. Indeed, Oracle Chairman Larry Ellison regularly claims, including again this year, that the cost to run Oracle databases on its Cloud is less than half that when running on AWS.
  • Openness and standards to facilitate multi-cloud operations and lower the barriers between migrating applications and data between clouds. Oracle cites its membership in the CNCF, which it expanded this year, the Cloud Security Alliance (CSA), Internet Society and voluminous contributions to various open source projects.
  • End-to-end security with a mix of hardware, operational processes and services, including several new ones just announced, to protect systems, data and users.

Indeed, in his opening keynote, Oracle founder Larry Ellison highlighted its strategy of accommodating existing applications, saying:

Our goal has always been as we move from one generation of computing to the next, to protect your investment in data and applications and make it easy to lift and move that into the next generation. We did that with Gen 2 Cloud.

Note, "Gen 2 Cloud" is Oracle's marketing slogan to differentiate itself from ostensibly first-generation products like AWS. Given that AWS, Azure and Google Cloud each introduce hundreds of new features and updates a year, it's unclear how one categorizes them as a particular generation since the products and variety of services offered today are vastly different than those of just a couple years ago.

Taking on AWS

At OpenWorld, Oracle again took dead aim at cloud market leader AWS by claiming its so-called second-generation cloud offers better performance, lower prices, “unprecedented availability” and more complete security. Each of these is highly debatable, however Oracle did reference benchmarking studies, some sponsored, some not, showing that its database services and raw compute infrastructure deliver higher performance at lower cost than comparable AWS offerings. While I’m not in a position to replicate the results, don’t be surprised to see AWS and other cloud competitors providing rebuttals soon.

One point AWS could legitimately contend about the benchmarking results highlighted by Oracle is the narrow focus on storage I/O performance using a particular type of bare metal compute service. Of course, bare metal offerings like the AWS i3.metal instances are designed with NVMe drives to provide a fast storage system, but Oracle has clearly done a good job of tuning its bare metal infrastructure for the high throughput, low latency requirements of its database. Nevertheless, they are but one of the scores of instance types AWS offers.

In a blog touting I/O performance 2- to 5-times that of AWS (depending on the use of local or remote storage respectively), product manager Andrew Reichman wrote that the difference is due to Oracle’s network design (emphasis added),

Oracle has a next-generation cloud network that connects our cloud components, including between servers and the block storage sub-systems. The network has no resource over-subscription, so performance doesn't get compromised when the network gets busy. Further, we used a flat network topology, which reduces the number of hops and the associated latency between any two devices. Off-box network virtualization offloads the effort from the server, which reduces the performance tax that customers would see without such an approach. Finally, storage traffic uses the full 25-Gbps pipe to the server, while AWS confines the storage traffic to their EBS optimized link that’s limited to 10 Gbps for their bare metal instance.”

Note, this last point is an exaggeration since the AWS i3 instance product page clearly states that they provide “up to 14 Gbps of dedicated bandwidth to Amazon Elastic Block Store (Amazon EBS),” but what’s 40 percent among friends?

Ellison directly targeted AWS in his keynote, which as Derek du Preez noted in his coverage, merely continues his history of sniping at competitors. Ellison claimed that Oracle cloud is:

  • 31% the price of AWS for compute instances per normalized CPU core
  • 2.5% the price of AWS for block storage with mid-range IOP performance
  • 1% the price of AWS for external data transfers (egress fees)

Original photo courtesy @matteastwood on Twitter

Ellison also stated that Oracle’s data warehouse and new Exadata-based transaction processing services are about an order of magnitude faster than native AWS alternatives Redshift and Aurora respectively.


Original photo from Derek du Preez

It’s difficult to dispute benchmarking claims since they are easily manipulated by cherry picking workloads and configurations that are favorable to one’s products. Nevertheless, in doing an earlier analysis of bare metal compute instances with similar configurations, I found Oracle’s service to be about 30 percent the price of the AWS i3.metal offering, so there are undoubtedly scenarios where Oracle Cloud is substantially cheaper than its biggest competitor.

Touting security

Security is a primary theme of Oracle’s cloud messaging at OpenWorld, where it announced four new services:

  • A Web application firewall (WAF) providing application layer (L7) filtering
  • A DDoS detection and migration service to defend against traffic flooding such as recent attacks using IoT-based botnets.
  • A key management service to handle encryption keys used for application and network access.
  • A Cloud Access Security Broker service to monitor cloud environments for adherence to security policies and detect anomalous behavior.

These are significant additions to Oracle's portfolio, however each of its primary competitors already offers similar capabilities. Indeed, the claim that "Oracle has built integrated layers of defense that are designed to secure users, apps, data and infrastructure," could truthfully be made by any of the other three mega cloud vendors.

Oracle's cloud transformation

Oracle's strategy of investment protection results in some confusion about how well its cloud business actually does since the company now consolidates cloud and on-premise revenue for both software licenses and support contracts. The logic behind this is Oracle's 'bring your own license' (BYOL) policy for cloud deployments that allows customers to migrate workloads from on-premise to Oracle Cloud without incurring additional software charges (of course, cloud deployments will incur charges for infrastructure services). As Ellison said during Oracle's Q1 2019 earnings call:

Many of our other customers, now that we have BYOL can move their Oracle database license from on-premise into our cloud, and they just pay basically for the incremental infrastructure cost. And they continue paying support on what becomes a cloud license.

Such an admixture of the much larger on-premise licensing revenue with a small, but growing cloud cohort has created confusion among financial analysts trying to untangle Oracle's reports. As one analyst put it in a question, "how do you get investors comfortable that this also is not obfuscating any cloud weakness?"

Co-CEO Safra Catz responded that the company had $1.7 billion in total cloud revenues in the latest quarter, i.e. about 19 percent of the total, and that new licenses deployed to the cloud are counted as cloud revenue. She later stated that:

Total cloud services and license support revenues for the quarter were 6.8 billion, up 5% in constant currency, 7% in US dollars.

Such cloud growth is a far cry from that reported by AWS and Azure, so investors are naturally looking for some leading indicators of cloud strength. Another analyst asked about customer demand for Oracle's showpiece product, the Autonomous Database, wondering where they are targeting new workloads: on-premise or the cloud. Ellison replied (emphasis added)

New workloads — all the new workloads for any given year is small compared to the 25, 30 years of inventory of data that we've accumulated since we've been in this business. So again, overwhelmingly existing databases, not new workloads, overwhelmingly BYOL as opposed to database services. … But you said, 'Where's the majority of it?' [ cloud deployments ] The majority of it is inside of our installed base. They're using their existing licenses. They're buying additional options. They're moving it to the autonomous database in the cloud.

Thus, most of Oracle's database deployments come from existing customers, which is almost every significant enterprise, moving on-premise Oracle databases to its cloud service. Given that reality, Oracle's emphasis on "investment protection" and enterprise robustness make sense since its customers treat Oracle more like an MSP than a pay-as-you-go cloud service.

My take

Like most established software companies, Oracle is in the midst of a strategic pivot from sole reliance on licensing and service revenue for on-premise deployments to subscriptions for cloud infrastructure and application services. Unlike AWS and Google Cloud, Oracle has focused on building an environment to accommodate legacy workloads and that minimizes the disruption in migrating database-backed applications from on-premise to cloud deployments. As such, its Cloud 2.0 looks more like Cloud 0.5: a great stepping stone from enterprise data centers to cloud managed services.

While Oracle spent ample keynote time discussing its infrastructure, its cloud portfolio includes a range of platform and application services. Indeed, I could see Oracle’s SaaS database and enterprise application products for CX, ERP, SCM, etc. becoming the preferred destination for organizations moving workloads to the cloud that don’t need or want the overhead of managing the underlying software and servers. Oracle is in a perfect position to follow the path of Adobe and Microsoft, which shifted office and content creation apps to a hybrid mix of locally-installed and cloud-based services.

Oracle rightly emphasizes the features and performance of its database services, however AWS, Azure and Google Cloud each have compelling database and data warehouse services like Redshift, SQL Data Warehouse and BigQuery respectively. While Oracle could (and should, given its long history in the space) have technical superiority for now, its competitors aren’t standing still and are aggressively innovating in database, data analysis and machine learning services as witnessed by Azure Cosmos DB and Google Cloud Spanner, massively scale out systems for which Oracle has no direct competitor.

An important metric of Oracle’s future cloud success, will be how rapidly it can shift the mix from license migrations to new workloads. Significant growth requires convincing customers to build around its cloud infrastructure, not just using it as outsourced infrastructure.