How Oracle's latest announcements change multi-tenant vs on-premises decisions
- Summary:
- Oracle’s announced some changes that give software buyers a new choice: a solution that provides the automated upgrades and performance updates that public cloud apps possess but on private cloud environments. While it mostly appeals to large enterprises, it is precisely those prospective customers that public cloud apps providers want to retain.
Oracle recently made a number of announcements regarding its autonomous database technology, its cloud Fusion applications and its Oracle Cloud Infrastructure (OCI). Mixed in these announcements were a number of new angles around its private cloud or cloud-in-a-box (CIAB) offerings, too.
While these announcements were interesting on their own, it’s what lies beneath Oracle’s new capabilities that is intriguing and more newsworthy. In short, Oracle’s now offering something that provides some of the best features of cloud, multi-tenancy (eg: vendor provided application maintenance), but in a highly secured private cloud/on-premises environment.
Let’s dig in…
The apps choices we used to have
To fully understand the weight of Oracle’s announcements, you need to first understand how companies have bought, implemented and maintained application software over the last few years.
For the last three decades, we’ve seen a number of changes in how software is sold, serviced and upgraded. One method, the on-premises style of computing, required customers to do it all. Buyers licensed or created applications that ran on computing hardware and systems software that the customer purchased and maintained. Some customers even acquired land and built special data centers with pricey power and cooling, backup generators and other components.
During the 1990’s, customers could acquire some applications via web applications. Many of these were niche solutions in the HR or CRM space, but their functional scope was limited at this time. Some of those vendors built their own data centers and upgraded the apps for their customers. Except for these few vendors, the on-premises deployment model was, overwhelmingly, the predominant deployment method. If anything changed in the 90’s, it was that many more on-premises applications became available on lower cost midrange and Unix servers.
But change did start in the late 90’s as we saw hosting firms emerge. Some of these companies provided hardware with single-tenant applications on them, while others only provided computer and networking hardware. Each customer still maintained their own data and applications while the hosting firm provided the equipment.
The following decade, the 2000’s, saw the arrival of multi-tenant cloud applications. These applications marked a major departure in prior computing deployment models. With multi-tenancy, the application software vendor acquired all of the needed computing hardware and systems software. All customers had to do was pay an annual subscription fee and the vendor took care of everything else including application maintenance.
For those vendors that couldn’t produce a multi-tenant version of their applications, many offered a single-tenant solution in a hosted environment. These were a poor cousin of multi-tenant solutions as the customer was still on the hook to provide all manner of support, upgrades and maintenance.
In the last decade, we saw hyperscalers emerge (eg: Amazon AWS). They offered a measure of scale few application software vendors could. Their data centers, global in scope, offered many times greater computing capabilities.
Through all of this evolution, we basically ended up with two kinds of solutions: those that customers must maintain themselves (ie: custom, on-premises, hosted and single-tenant apps) and those that vendors maintain (the multi-tenant cloud apps).
Well, that’s changing.
Oracle’s latest news
Oracle’s Autonomous Database (Autonomous DB) was originally designed to run on Oracle’s public cloud (OCI – Oracle Cloud Infrastructure). It’s a set of capabilities wherein Oracle continuously secures, tunes and updates the Oracle Relational Database that customers are utilizing. The Autonomous DB can be beneficial as it transfers much of the (tedious) systems management activity to Oracle. It also provides a very proactive level for security.
Oracle also automated many of the support functions within the Autonomous DB. Oracle uses a number of tools to automatically patch its products including artificial intelligence utilities to detect security and performance problems. The speed with which Oracle can detect and repair/tune these solutions may be much faster than what a customer’s IT department can do.
Autonomous DB
Oracle’s recent announcements focused on Oracle’s Autonomous Database and its new availability in private cloud environments. With that new availability, Oracle’s Fusion Cloud apps can take advantage of the autonomous DB environment in private cloud/cloud-in-a-box environments. Some of the new capabilities are targeted to larger firms in specific industries where private clouds are common.
Underlying all of Oracle’s announcements is a single observation: Oracle can offer its applications (that it will maintain for clients) on hardware, systems software and database technology that Oracle will optimize automatically for clients. These capabilities will exist on its public cloud Oracle Cloud Infrastructure (OCI) and on private clouds. What this means is that Oracle now has solutions that scale up and down in a highly efficient and timely manner over two kinds of technology environments. Customers will only use enough disk storage, memory, etc. that is required in these environments without overbuying compute capacity.
Oracle’s solution will scale automatically at peak times. Oracle applications, databases and other solution components stay current, get patched, and are tuned and optimized via these combined capabilities. What Oracle will do automatically, other solutions would require a customer to complete on their own and possibly via multiple vendors (eg: a hyperscaler, a DBMS vendor, a virtual machine solution provider, an application software vendor, etc.). Dealing with so many providers is inefficient and exposes customers to risk unless they overbuy capacity.
The specific value proposition of the Autonomous DB is that it helps customers avoid the expense: of hiring database administrators (to tune and maintain the RDBMS environment); of having developers spend time on low-value-add activities (eg: securing and patching the systems software); and, losing time in provisioning new databases. According to Oracle, automated patching and always-on security - such as identifying potential security breaches and remediating these threats - are key drivers for customers.
Now Oracle is making the Autonomous DB available to customers using Oracle’s on-premises Exadata hardware (Exadata Cloud @ Customer). This takes the Autonomous DB beyond the public cloud OCI world and into other market segments including customers’ on-premises data centers and private cloud environments.
Oracle also announced that it has now certified its on-premises apps from Siebel, PeopleSoft and JD Edwards will also work with the Autonomous DB.
The economics of dedicated Region Cloud @ Customer
One application of these new capabilities is Dedicated Region Cloud @ Customer. This is a solution targeting very large customers who want a cloud technology environment on a pay-as-you-use model, but on hardware they acquire from Oracle to meet data residency or performance requirements. Oracle is placing all of its public cloud services and applications into a customer’s on-premises data centers. This solution resides on Oracle’s cloud-in-a-box (CIAB)—Cloud @ Customer—hardware/software solution.
This solution targets a specific ICP (ideal customer profile). Initial pricing starts at $500,000/month with a 3-yr minimum commitment. The customer gets a copy of Oracle’s OCI environment running on their CIAB hardware plus any of Oracle Fusion SaaS applications.That $18 million cost over 3 years is an operating expense, not a capital expense.
Companies with significant data sovereignty and security concerns are the targets for this offering. This would likely include companies in the financial services vertical, regulated industries and firms with significant national defense contracts. It could also appeal to firms trying to transition to a hybrid computing environment.
Competitive implications
There are some interesting competitive implications of this solution. Oracle is targeting a specific demographic that may not want a public cloud solution. These firms are some of the most desirable customers any application software would want. And, Oracle is positioning itself to provide its cloud applications on its own public cloud or on a customer’s private cloud via the Cloud@Customer deployment option. (Readers might also want to evaluate AWS Outposts and the Microsoft Azure Stack options.)
Oracle’s application software competitors generally offer only one solution deployment option (public cloud or on-premises). Very few application software vendors can offer a choice of on-premises or public cloud and, if they do, it comes with some limitations.The most notable limitation is that an application running on-premises may not get the automatic upgrades/patches/etc. that their public cloud solution gets. Oracle now offers this regardless of the deployment environment.
Oracle’s solution has some other strengths. Most notable is that Oracle controls the entire stack from hardware to applications and everything in between. Because all customers for this will have the same configuration (and the only one Oracle will need to support), it means Oracle can support all of its customers with speed and low cost. This will be true whether the customer is using the Cloud@Customer or OCI public cloud deployment.
Competitors can only control one environment - their public cloud environment. They can’t necessarily control the hardware and systems software that customers might possess in their own private cloud solutions. That lack of control makes provisioning of multi-tenancy more difficult and/or more costly for the private cloud versions of these solutions. These competitors can create a standalone private cloud for a major customer and run a single-tenant instance of their public cloud solution there. It’s an expensive, one-off approach that hurts the economics of their multi-tenant public cloud solutions. It also doesn’t scale well.
Oracle’s offering will trouble some of its competitors. My experience with really large enterprises tells me that they want:
- To control when applications get upgraded. Some clients are really frustrated that public cloud application solutions have a pre-defined and frequent upgrade schedule. While most firms are delighted to get software upgrades multiple times per year (as opposed to the once every 18-24 months with old on-premises apps), large firms are trying to keep a huge, global portfolio of applications running with zero disruptions to their far-flung business empire. This is why upgrade control is so important to these companies. While large firms want to control the timing of upgrades, they don’t want to forgo them altogether. Subscribers to Oracle Cloud services get automatic updates. The customer doesn’t do any updates when it comes to Oracle Cloud.
- To keep their information on machines they control. This is often an emotional issue as most cloud applications physically and logically separate each customer’s data from all others. Oracle also offers container technology to further separate and secure customer data. And while some other cloud apps vendors use containers, there are some large enterprises that will state, unequivocally, that customer commitments, regulations or other requirements dictate that they can’t place their customer data on a public cloud.
Apps vendors that only offer a public cloud version will find some large enterprises less interested in their offerings as they may want the cloud software running locally on their own hardware. Or, these larger firms will want the flexibility to move apps from public cloud to private cloud whenever they choose.
Interestingly, I do know of two major cloud apps vendors that have created a couple of one-off private cloud versions of their product line for a major customer or two. But, these deals are very confidential, one-offs and not an advertised solution they offer. And the reason for this is that they can’t provide timely application updates to customers with dissimilar technical architectures.They also can’t support these environments as the customer will likely lock them out of these data centers. Why? These customers won’t want anyone to have access to their proprietary information or their customers’ data.
I’m also aware of some apps vendors who sell an older on-premises version of their product in a hosted manner. By hosting an older solution on a public cloud or hyperscaler platform, these vendors will claim they are providing both a public cloud and a private cloud version of their apps. However, most buyers see this re-packaging for what it is. It’s an old product being tarted up and not a real cloud-first application solution.
A big challenge for Oracle competitors is that they do not possess the entire stack. They don’t have their own hardware, systems software, etc. Yes, some of them use open-source tools for some systems software components and some might specify a given hyperscaler for compute power, but this environment is not a ‘one throat to choke’ world. If something’s amiss with Oracle’s solution, Oracle must fix it. If something fails in a competitor’s environment, it could be the application software vendor’s fault, some bad open source code, a hardware issue, a bad systems software patch, a hyperscaler issue, etc. Customers must choose between distributed, diverse accountability and single source accountability.
My take
Oracle’s providing another deployment option for enterprise application software. It clearly rests between the world of on-premises and public cloud/multi-tenant solutions. It offers the benefits of multi-tenant apps but in a computing environment that is not shared with other customers. Oracle controls the infrastructure while the customer controls the data. It clearly represents a better choice than an old on-premises solution where the customer (not the vendor) had to maintain the hardware, systems software, application software, etc.
Traditional Oracle applications competitors (eg: Workday, SAP, Salesforce, etc.) may find this single stack offering hard to best and a real challenge in sales situations. They may have to offer more flexibility in the timing of their software upgrades and/or permit customers to run their applications on different computing stacks with all of the support issues that entails. I’d expect these firms to offer their solutions on a few hyperscaler environments (eg: Google Cloud Platform) where the vendor can run a single customer version of the solution on a standardized technical environment. That approach would let the vendor avoid a lot of different stacks to support. Will competitors do it? I believe they already are – just look at their ITAR solutions for government contractors. Will they expand this? Maybe for some very large customers only. We should expect a very measured approach as vendors lose economies of scale with every unique customer environment they must support.
I asked Oracle executives if they’d alter their contracts to make every aspect of software usage (and the attendant financial considerations) also scalable as only they are in a position to make this a competitive advantage. While other ERP vendors can scale their software up or down on their own public cloud (or third party hyperscaler) machines, those solutions are only available on public clouds and/or require the customer to overbuy subscriptions to handle peak demand. Oracle controls the entire stack and could permit customers to easily ramp up or down their consumption to match their changing business situation. That would have been a welcome capability for a number of firms whose businesses got decimated in this pandemic. Let’s see if they consider this suggestion.