Shining a light on Oracle Cloud

Grazed from InfoWorld. Author: Eric Knorr.

It’s been over a month since Larry Ellison strutted across the stage and unveiled Oracle Cloud and its first three enterprise applications: CRM, human capital management, and enterprise social networking. Plus, Ellison took the wraps off cloud versions of WebLogic and Oracle Database itself.

All told, Oracle’s fearless leader said that more than 100 applications would be available, including ERP eventually — typically the last application category enterprise customers consider trusting to the cloud.

Quite a few responses to the announcement have been cynical: another golden opportunity for Oracle lock-in. Or: The same old stuff, now available by subscription through the cloud at rates Oracle hasn’t even seen fit to announce yet. Some even questioned whether it was a cloud offering at all, since Oracle touted the fact that each customer would get its own instance of the software…

But it’s too early to pass judgment, precisely because so many questions about Oracle Cloud remain. At this early stage, you won’t find many customers willing to talk, since Oracle is onboarding them one by one — basically, you register and get in the queue.

To gain more insight into Oracle Cloud, InfoWorld Executive Editor Doug Dineley and I spent an hour interviewing Abhay Parasnis, senior vice president of Oracle Cloud. First of all, we wondered, why the highly controlled rollout?

Oracle won’t disclose the current number of customers, but Parasnis said this was not because the applications themselves were unfinished. He gave a different explanation:

A lot of the customer base that is interested, as you can imagine, is the enterprise-class customer base, which is used to a certain level of stability and platform robustness when dealing with Oracle. Not to diminish some of the other players in the market, but they’re not doing this as "let’s just try out a toy application somewhere." I think the expectation, regardless of how slow or fast they start, is an enterprise-grade platform.

This cautious approach indicates perhaps the biggest difference between Oracle Cloud and other SaaS and PaaS (platform as a service) offerings. Salesforce.com, for example, has been proud of "sneaking" its SaaS apps into the enterprise; lines of business decide they can’t get what they want from IT and turn to a SaaS provider instead. By contrast, Oracle seems intent on walking in the front door and selling to upper enterprise management of existing Oracle customers.

This focus on enterprises needs extends to the way Oracle will push out upgrades. As Parasnis explained:

When there comes a time when the vendor is ready to either upgrade or release new functionality, there are two broad schools of thought in the market today. For you to get the benefits of SaaS, where somebody else is going to manage your application, you must sign up to give up control in terms of when and how and what cadence you will actually upgrade your users to the new version, new workflows, or new capabilities. For some companies, or for some applications, that may be acceptable. For most enterprise users, ideally what they want are the benefits that cloud upholds in terms of somebody else managing the footprint and the lifecycle of applications, hopefully a lower cost structure. But more importantly, they don’t want to lose the control.

To accommodate that desire for control, said Parasnis, Oracle will provide upgrade windows and work with customers to determine a convenient time to make the transition to new functionality or new versions. Now, of course, this is possible only because Oracle Cloud applications are not multitenanted. Instead of one huge instance of an application, to which all customers subscribe, Oracle deploys one instance per customer, which from a security standpoint supports greater isolation among customer data stores as well.

Single tenancy has led many people, including me, to wonder whether Oracle Cloud is a cloud at all. On the surface, in an architectural sense, it sounds a lot more like conventional hosting. So is it? Here’s Parasnis’ reply:

In the cloud, there are a few things in my mind that really distinguish what we are doing from "hosting." First of all, the core underpinning of the entire cloud is in fact this elastic capacity model. We are not literally trucking in servers and starting apps on those servers and offering those copy by copy to customers. There is an elastic substrate, all modeled as a set of virtual machines, running on either Exadata or Exalogic depending on the workload … and a combination of virtualized network and virtualized compute and virtualized storage. We actually think about it as a logical fabric that is fully orchestrated and managed like any "full first class" cloud.

On top of that, we have taken the exact same philosophy and extended the Fusion apps themselves. The core functionality of the apps is exactly the same, but the app servers that they are running on and the application architecture itself is now deeply wired into this cloud substrate.

Basically, everything underneath Fusion apps is pretty darn cloudy, even if the apps themselves are not multitenanted. It’s legitimate to question how Oracle is going to be able to ensure a seamless customer experience for all those single tenants. But if it all works the way Parasnis says it will, the result could be more attractive for some enterprise customers than a "pure" SaaS solution:

Frankly, because the stacks [in the cloud and on premises] are actually the same, we think there are a few things that are very difficult for competitors to match. First of all, there are skills and investment portability. There is a lot of customization and a lot of investment customers have made on top of our applications and on top of our platform on premises. We can give them a natural progression and natural path either to move them — but in quite a few cases not really move but connect to those on-premises extensions with some new things they are starting to do on the cloud.

That may be the bottom line. Large enterprises invest heavily in their customizations and workflows, and in "moving to the cloud," many may prefer a solution like Oracle’s that purports to get them there with as little disruption as possible. By the same token, Oracle has invested billions in developing its applications, so if possible it’s going to avoid re-engineering them for multitenancy — and sell single tenancy as a security benefit.

Besides, who really knows what goes on behind the scenes at "pure" SaaS providers? Do all of them really have one giant application instance for all customers, or do some actually have an architecture similar to Oracle’s? If the customer experience meets expectations, nobody cares.

The most significant thing about Oracle Cloud is that Oracle felt compelled to create it — and that it specifically targets the enterprise. Never underestimate the power of the Oracle marketing machine to reach those customers, although adoption is likely to be slow and cost is a huge question. If nothing else, Oracle Cloud is a sign that the cloud had grown up [6] — all the way up to the biggest enterprise apps and customers.